TECHNICAL GUIDE Responsible Digital Credit Frontier Solutions for Authorities and Providers July 2025 • Juan Carlos Izaguirre, Sonia Arenaza, Patrick Meagher, and Myra Valenzuela Acknowledgments Rights and Permissions The authors would like to thank CGAP colleagues This work is available under the Creative Commons Haocong Ren and Eric Duflos for extensive input Attribution 4.0 International Public License (https:// and guidance, Bryce Feibel for research support, and creativecommons.org/licenses/by/4.0/). Under the Nokuthula Nkhoma for editorial support. We are also Creative Commons Attribution license, you are free grateful to peer reviewers Silvia Baur-Yazbeck, Emilio to copy, distribute, transmit, and adapt this work, Hernandez (CGAP), and the members of our technical including for commercial purposes, under the following advisory group: Paul Adams (IPA), Patrick Conteh (Africa conditions: Fintech Network), Ivor Istuk (World Bank), Philip Rowan (CCAF), Sheila Senfuma, Stefan Hall (Consumers Attribution—Cite the work as follows: Izaguirre, Juan International), Matthew Soursourian (OECD/FinCoNet), Carlos, Sonia Arenaza, Patrick Meagher, and Myra and Jayshree Venkatesan (CFI). Additionally, we thank Valenzuela. 2025. Responsible Digital Credit: Frontier Cristina Martinez and Gerhardus Coetzee for their initial Solutions for Authorities and Providers. Technical contributions to this work, as well as participants of the Guide. Washington, D.C.: CGAP. https://www.cgap. Responsible Finance Forum in 2023, 2024, and 2025 org/research/publication/responsible-digital-credit- who provided useful feedback to this work.. solutions-for-authorities-and-providers Translations—If you create a translation of this work, CGAP add the following disclaimer along with the attribution: This translation was not created by CGAP/World Bank 1818 H Street, NW, MSN F3K-306 and should not be considered an official translation. Washington, DC 20433 CGAP/ World Bank shall not be liable for any content or Website: www.cgap.org error in this translation. Email: cgap@worldbank.org Telephone: +1 202 473 9594 Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the Cover photo by Sujan Sarkar | India. attribution: This is an adaptation of an original work by CGAP/World Bank. Views and opinions expressed in © CGAP/World Bank, 2025. 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. Responsible Digital Credit: Frontier Solutions for Authorities and Providers ii Contents Executive Summary 1 SECTION I: Introduction 4 Understanding Personal Digital Credit 5 The Growth and Transformation of Digital Credit 6 Personal Digital Credit Brings Important Benefits to Borrowers 7 Behavioral Vulnerabilities Associated with Digital Credit 8 New and Evolving Consumer Risks in Personal Digital Credit Could Lead to Negative Outcomes 8 A Responsible Ecosystem Approach Is Needed for Borrowers to Fully Realize the Benefits of Digital Credit 11 SECTION II:Consumer Risks in Personal Digital Credit: A Customer-Centric View 12 Customer Journey in Personal Digital Credit 13 Provider Life Cycle in Digital Credit 17 The Fragmented Regulatory and Institutional Arrangements in Digital Credit 20 SECTION III: Frontier Solutions For Responsible Personal Digital Credit: Authorities 23 Responsible Digital Credit Solutions by Authorities 25 A Call to Action for Supervisors, Regulators, and Policy Makers 45 A Call to Action for Funders 45 SECTION IV: Frontier Solutions for Responsible Personal Digital Credit: Providers 47 Responsible Digital Credit Solutions by Providers 49 A Call to Action for Digital Lenders and Service Providers 65 A Call to Action for Funders 66 SECTION V:  rontier Solutions for Responsible Personal Digital Credit: F Research Organizations and Consumer Representatives 68 Role of Research Organizations in Promoting Responsible Digital Credit 68 Responsible Digital Credit Solutions by Research Organizations 69 Role of Consumer Representatives in Promoting Responsible Digital Credit 75 Responsible Digital Credit Solutions by Consumer Representatives 75 A Call to Action for Consumer Representatives and Research Organizations 78 A Call to Action for Funders 78 APPENDIX I: Selected Examples of Solutions by Authorities 79 APPENDIX II: Selected Examples of Solutions by Providers 88 References 93 Responsible Digital Credit: Frontier Solutions for Authorities and Providers iii Acronyms and Abbreviations AES Advanced Encryption Standard AFI Alliance for Financial Inclusion AI Artificial Intelligence API Application Programming Interface APP Authorized Push Payment APR Annual Percentage Rate ASIC Australian Securities and Investments Commission AWS Amazon Web Services Bigtech Big Technology Company BNPL Buy Now, Pay Later CAK Competition Authority of Kenya CBK Central Bank of Kenya CCAF Cambridge Centre for Alternative Finance CEGA Center for Effective Global Action CFI Center for Financial Inclusion CFPB Consumer Financial Protection Bureau (US) DFS Digital Financial Services ECOA Equal Credit Opportunity Act (US) EMDE Emerging Market and Developing Economy EU European Union FAQ Frequently Asked Question FCA Financial Conduct Authority (UK) FCCPC Federal Competition and Consumer Protection Commission (Nigeria) FinCoNet International Financial Consumer Protection Organisation Fintech Financial Technology Company FIU Financial Intelligence Unit FSA Financial Sector Authority FSD Financial Sector Deepening FSP Financial Services Provider FY Fiscal Year GDPR General Data Protection Regulation (EU) HKMA Hong Kong Monetary Authority ICT Information and Communications Technology IPA Innovations for Poverty Action KES Kenyan Shillings KYC Know Your Customer MFA Multifactor Authentication MFI Microfinance Institution Responsible Digital Credit: Frontier Solutions for Authorities and Providers iv ML Machine Learning MNO Mobile Network Operator MSE Micro and Small Enterprises NCBA National Commercial Bank of Africa NLP Natural Language Processing NPC National Privacy Commission (the Philippines) ODR Online Dispute Resolution OECD Organisation for Economic Co-operation and Development PIN Personal Identification Number QR Quick Response RBI Reserve Bank of India RBIH Reserve Bank Innovation Hub (India) RDFE Responsible Digital Finance Ecosystem Regtech Regulatory Technology ROSCA Rotating Savings and Credit Association SIM Subscriber Identity Module SMS Short Message Service T&C Terms and Conditions TLS Transport Layer Security UK United Kingdom UMRA Uganda Microfinance Regulatory Authority US United States of America USD US Dollars USSD Unstructured Supplementary Service Data VSLA Village Savings and Loans Association Responsible Digital Credit: Frontier Solutions for Authorities and Providers v Executive Summary T HE DIGITAL CREDIT MARKET IS RAPIDLY authorities, providers of digital loans and digital credit evolving, shaped by new players, services, and research and consumer organizations, technologies, and partnerships. While this the guide categorizes solutions across four phases evolution brings financial, experiential, and welfare of consumer risk management: identification, benefits to borrowers, it also introduces heightened prevention, mitigation, and resolution. consumer risks. Financial Services Providers (FSPs) may exploit behavioral biases, exposing consumers to While the guide offers practical solutions for authorities fraud, data misuse, lack of transparency, inadequate and providers, it also includes a call to action for funders, redress mechanisms, and unfair treatment. These risks, research organizations, and consumer representatives, particularly in contexts with fragmented consumer recognizing their critical role in shaping a responsible protection frameworks, can lead to overindebtedness digital credit ecosystem. As the digital credit market and deteriorating financial health. continues to evolve, this guide does not evaluate the effectiveness of each initiative. Instead, it presents This calls for a responsible ecosystem approach, one a typology of emerging practices, highlighting key in which authorities, providers, and other stakeholders benefits and limitations to support stakeholders in collaborate to establish customer-centric rules and tailoring solutions to their specific markets. practices and strengthen their commitment and capability to mitigate and address consumer risks Our research shows that authorities tend to (Duflos et al. 2024). Customer centricity involves: prioritize preventative solutions, particularly in addressing fraud, transparency, unfair treatment, • Understanding the entire borrower journey, from and overindebtedness. Authorities are laying the loan application to repayment and redress. groundwork for a more responsible digital credit market • Mapping the provider lifecycle behind the by registering and licensing digital lenders, and by borrower journey, including technology-driven and collaborating across agencies on regulatory, market partnership-based stages. monitoring, and market intelligence-sharing efforts, • Clarifying the financial regulatory perimeter of particularly to identify and prevent fraud. Regulatory digital lenders and digital credit service providers, measures such as improved disclosure requirements, the including the roles of financial and non-financial use of positive friction, and rules on product governance authorities. and algorithmic credit scoring enhance transparency and fairness. To mitigate and resolve conduct risks, Following a customer-centric methodology, this authorities are introducing initiatives including fraud technical guide presents practical solutions that compensation, personal insolvency, anti-predatory authorities and providers can adopt to make personal lending regulations, and responsible debt collection. digital credit more responsible. Drawing from a global Strengthened collaboration with ecosystem actors landscaping of initiatives by financial and nonfinancial Responsible Digital Credit: Frontier Solutions for Authorities and Providers 1 will be essential to developing more customer-centric regulations and improving market monitoring. Summary Results of the Global Landscaping Providers are leveraging technology to address risks, This desk research identified 160 initiatives aimed at especially fraud, data misuse, and overindebtedness, addressing the various risks faced by personal digital more comprehensively. Tools such as tech-enabled credit borrowers. Numerous initiatives span multiple authentication, data analytics, fraud prevention and risks, phases of the risk management cycle, and detection, and customer support systems help to stages of the customer journey. The majority were led protect consumers from financial harm. Positive by government authorities (see Figure 1). For these friction, dynamic pricing, and automated debt solutions, our unit of analysis was a specific regulatory monitoring support more responsible lending practices. chapter or clause, rather than an entire regulation or Privacy enhancing features like data encryption, guideline. As a result, a single digital credit guideline or sharing preferences, and consent management tools regulation may have been disaggregated into multiple help reduce data misuse. Across the ecosystem, initiatives, each linked to different consumer risks. collaborations among lenders, big technology companies (bigtechs), and telecommunications companies, and other service providers are increasingly FIGURE 1. Initiatives per actor essential to address consumer risks across the product 60% lifecycle. We urge providers to continue to leverage technology and collaboration to enhance consumer 51% security, product suitability, empowerment, and overall 50% trust in digital credit. 40% Researchers, consumer representatives, and funders play critical roles in advancing a responsible 30% digital credit ecosystem. Research and consumer organizations have partnered with authorities and 18% providers to monitor borrower risks, generate market 20% insights, and test and advocate for solutions. We 12% 10% encourage them to deepen these efforts and strengthen 10% 5% 5% their ability to engage effectively with authorities and providers. We also encourage funders to disseminate 0% and advocate for solutions, conduct thorough due Solutions by actor diligence, monitor the adoption of solutions, and strengthen collaboration among stakeholders. Authorities 3rd party tech providers Ultimately, only collaborative efforts Financial services providers will lead to a responsible digital credit Bigtechs ecosystem. Research organizations Consumer & industry associations Source: Authors. Note: Values do not add to 100 percent due to rounding. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 2 Regarding the customer journey, most initiatives focus on the steps of application, repayment, and agreement stages (see Table 1). In terms of consumer risks, the most commonly addressed issue was lack of transparency, followed by fraud, data misuse, and unfair treatment (see Table 2). Across the risk management cycle, the majority of initiatives targeted risk prevention (108) and risk mitigation (63) (see Table 3). Initiatives by step in the TABLE 1.  Initiatives by consumer TABLE 2.  Initiatives by risk TABLE 3.  customer journey risk and issue management phase Customer No. of Initiatives Consumer Risk No. of Initiatives Risk No. of Initiatives Journey Step (Many Cover and Issue (Many Cover Management (Many Cover Multiple Steps) Multiple Risks) Phase Multiple Phases) Awareness/ 66 Behavioral 24 Risk Identification 52 Market Vulnerabilities Information Risk Prevention 108 Fraud 41 Advice/ 59 Risk Mitigation 63 Orientation Data Misuse 36 Risk Resolution 11 Account Sign-up 79 Lack of 48 Transparency Application 90 Source: Authors. Inadequate 19 Agreement 83 Redress Assessment/ 72 Unfair Treatment 33 Approval Agent-Related 11 Disbursement 63 Risk Using Funds 51 Network 19 Downtime Repayment 88 Overindebtedness 32 Closure 47 Late Repayment 56 Source: Authors. Recourse/Query/ 72 Grievance Source: Authors. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 3 Section I Section II Section III Section IV Section V Appendix I Appendix II SECTION I Introduction C ONCERNS ABOUT EVOLVING associated with a broad range of digital credit products consumer risks in digital credit are growing, (see Table 4). yet most research has focused on single case studies or pilots, rather than a comprehensive view of This Guide builds on CGAP research, and work from other solutions addressing these risks. This Technical Guide key global bodies, including the Alliance for Financial aims to fill this knowledge gap by: Inclusion (AFI), the International Financial Consumer Protection Organisation (FinCoNet), the Organisation for • Compiling evidence on how authorities, providers, Economic Co-operation and Development (OECD), and and other stakeholders across jurisdictions are the World Bank, as well as research organizations such working to prevent, identify, mitigate, and resolve as Innovations for Poverty Action (IPA), the Cambridge consumer risks in personal digital credit Centre for Alternative Finance (CCAF), and the Center for • Increasing awareness and understanding of the Financial Inclusion (CFI).  types of solutions being adopted, their relevance, associated challenges and limitations, and, where It also draws on CGAP’s decade-long work in digital available, their outcomes. credit, including gender-sensitive market monitoring with authorities and lab experiments with providers, To do this, we conducted a comprehensive desk which highlighted persistent challenges such as high research effort from April 2024 to March 2025, nonrepayment rates, limited transparency, inadequate drawing on publicly available sources, including recourse, gender disparities in credit access, fraud, and research publications, regulations, guidance, and news data misuse (CGAP 2025b; Izaguirre et al. 2022a; Mazer articles. This research identified 160 initiatives and and McKee 2018). systematically categorized them by: • The consumer risk addressed, This work contributes to CGAP’s vision for a Responsible Digital Finance Ecosystem (RDFE), where • The phase of the risk management cycle involved, financial and nonfinancial authorities, providers, and consumer representatives, and market facilitators • The type of actor leading the initiative. collaborate to ensure that digital financial services (DFS) improve people’s lives and protect them from This categorization provides a structured view of how financial harm (Duflos et al. 2024). The RDFE framework different stakeholders are responding to emerging risks is built on four components: in a digital credit ecosystem. While the compilation is • Customer Centricity: Placing customers at the extensive, it is not exhaustive. It reflects a diverse array center of every action. of initiatives adopted across different market contexts and varying maturity levels to address consumer risks • Collaboration: Jointly creating, funding, implementing, and monitoring solutions. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 4 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 4. Summary of key attributes Key Attribute Description Instant Loans are approved instantly, often within seconds. Automated Loan decisions are automated (no human review) and leverage alternative data (e.g., social media, behavioral data). Remote Account sign-up and application, disbursement, repayment, and queries are managed remotely, and often available 24/7. Direct to individuals Credit risk based on individual; initially reliant on alternative data, then shifts to conventional data (e.g., loan repayment). Unsecured Generally, no collateral, guarantee, cosigners, or formal proof of income required. Shorter term Loan tenure can be as short as 1 week, typically 4 weeks up to a few months. Smaller value First-time loans can be tiny (as low as $0.50) and then increase in value upon successful repayment. Sources: Authors, based on AFI 2020, CEGA n.d, Hwang and Tellez 2016, and Stewart et al. 2018. • Capability: Strengthening tools, skills, and credit products, i.e., credit offered to individuals institutional capacity. rather than micro and small enterprises (MSEs). Credit designed for MSEs generally requires distinct • Commitment: Demonstrating sustained leadership creditworthiness assessments and underwriting and accountability. processes, such as assessment of inventories, cash flows, personal activity of applicants, digital receipts, Understanding Personal Digital and physical assets (CGAP 2019).1 Credit This research also includes solutions targeting digitally Since their emergence in the early 2010s, digital loans delivered retail Buy Now, Pay Later (BNPL) products. have introduced new lending dynamics distinct from BNPL is a form of point-of-sale credit that allows conventional credit. CGAP initially identified three key consumers to defer payment for purchases and repay attributes that differentiate them from conventional interest-free installments within a set period. As BNPL loans: instant, automated, and remote (Hwang and Tellez products have become increasingly digitized, they 2016). More recently, Accion, AFI, the Center for Effective now share many of the same characteristics and pose Global Action (CEGA), and others have expanded this similar consumer risks to other forms of digital credit. and identified additional important attributes (AFI 2020; CEGA n.d.; Stewart et al. 2018), as shown in Table 4. Table 5 highlights several well-known examples of personal digital credit products operating in Ghana, While we recognize that money is fungible and that India, Kenya, Mexico, Nigeria, Peru, the Philippines, borrowers may use digital credit for entrepreneurial Tanzania, and Uganda. purposes, this research focuses on personal digital 1 In this Technical Guide, we use “consumers” to refer to the broader public and “customers” to refer to those who have active relationships with a provider. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 5 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 5. Examples of personal digital credit products and their main characteristics Provider Lending Category Partnership Launch Loan Loan Customers Markets Product Year Size Term (USD) Tala Tala Mobile- MNOs, banks 2011 10– 21–90 6 million India, based 1,000 days Kenya, Peru, Philippines, Tanzania Branch Branch Mobile- MNOs, banks 2015 2.2– 1–12 4 million India, Kenya, based 1,200 months Nigeria, Tanzania Safaricom M-Shwari Mobile- MNO 2012 0.78– Up to 6.5 million Kenya and (part of the based (Safaricom), 7,751 30 days (FY24) NCBA M-Pesa bank (NCBA) ecosystem) M-Kopa M-Kopa Mobile- MNO 2011 0.70– Up to 5 million Ghana, solar (pay- based and (Safaricom, 350 30 days Kenya, as-you-go) online Vodacom, Nigeria, MTN), bank Uganda (NCBA) Kubo Kubo Mobile- Financial 2012 1,200– 4–36 1,200 (FY24) Mexico Financiero based and institutions  5,000 months online Sources: Provider websites. The Growth and Transformation fraud detection, KYC, credit scoring). As partnerships increase, modularization is expected to shape the market of Digital Credit and the borrower journey. The global digital credit market has experienced rapid transformation, driven by technological innovation, Alternative Data and Embedded Finance: Digital evolving partnerships, and expanding data ecosystems. lending relies on vast data sources, including cell This evolution brings new opportunities and benefits, phone usage, social media activity, and payment and alongside new risks and challenges, for borrowers. transaction history (CGAP 2025a). Consumers, often willing to share data for credit access, may also do so Market Growth: The market has rapidly grown in inadvertently via financial services that are embedded the past decade. In 2019, fintech and bigtech credit in nonfinancial companies’ services (Fernandez Vidal reached $800 billion globally, with rapid uptake in and Salman 2023; Kruijff et al. 2024). Alternative Africa, Asia, and Latin America driven by cell phone data-based credit scoring supplements traditional penetration (see Figure 2). The global fintech credit scoring, enabling access for those without credit market is projected to reach $4.9 trillion by 2030 (Allied histories. Open finance facilitates data sharing across Market Research 2021). financial actors (CGAP et al. 2024). Ecosystem Complexity and Modularization: A growing AI/ML Risks and Benefits: AI/ML improves credit set of actors is reshaping the digital credit landscape, assessment, underwriting, pricing, and fraud detection, including fintechs, alternative data providers, mobile but also poses risks to consumers, including bias, network operators (MNOs), and providers of artificial discrimination, system disruption, or potential intelligence/machine learning (AI/ML)–driven tools (e.g., exploitation for fraud. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 6 Section I Section II Section III Section IV Section V Appendix I Appendix II FIGURE 2. Bigtech credit is overtaking fintech credit Consumption Smoothing: Digital credit provides short- term financing that helps borrowers meet household 900 needs and manage micro and small businesses. For example, in Tanzania, women were more likely to use 750 digital credit for business than men, while men tended to use it for household needs and airtime purchases 600 (Kaffenberger et al. 2018). USD bn Improved Customer Experience: Borrowers benefit 450 from lower transaction costs and greater convenience due to near-instant loan approval and disbursement. In 300 addition, using private digital channels allows borrowers to avoid interpersonal issues more common in other 150 borrowing environments, such as harassment, social pressure, and corruption (Mazer and Fiorillo 2015). 0 2013 2014 2015 2016 2017 2018 2019 Potentially Improved Personal Resilience and Well-Being: In Kenya, households using M-Shwari Lending volume: Fintech were 6.3 percent less likely to forgo expenses when Lending volume: Bigtech facing unexpected income shocks, suggesting improved short-term financial resilience (Suri et al. Source: Cornelli et al. 2020. 2021). In Nigeria, users of a digital lending app reported Note: The 2019 fintech lending volume figures are estimated for Australia, China, the EU, Great Britain, New Zealand, and the US. improved well-being, though the product did not lead to resilience or women’s economic empowerment (Björkegren et al. 2021). Public Goods and Governments’ Role: Government- led public goods, such as cybersecurity frameworks, interoperability regimes, and telco infrastructure, have been essential for building a trusted digital credit market. Customer Quotes on Positive Experiences BOX 1.  with Digital Credit “M-Shwari is so convenient for me. It’s a direct Personal Digital Credit Brings process, there is no waiting, and there is no holding Important Benefits to Borrowers you. It’s instant.” —Helen, Kenya (Cook and McKay 2015). The evolution of personal digital credit has brought “When I got approved for a [Tala] loan it was like important benefits to potential and existing borrowers: a huge weight has been lifted off my shoulders because I don’t need to worry about my water and “Gateway” to Formal Financial Inclusion: Digital electricity bills.” lenders increasingly use alternative data to assess —Jean, Philippines (Tala n.d.). creditworthiness, helping overcome barriers such “MoMo Kash is about helping others. When you’re as lack of collateral or formal credit history. This having trouble managing, you can use the service is especially valuable for women and low-income with a view to repaying within a certain timeframe.” populations, enabling them to access tailored financial —Anonymous male, Côte d’Ivoire (Navarro et al. 2024) products that better meet their needs (Caire and Fernandez Vidal 2024). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 7 Section I Section II Section III Section IV Section V Appendix I Appendix II Behavioral Vulnerabilities Salience Bias: The lack of human interaction and Associated with Digital Credit tangible processes can make digital credit feel less “real”. This may result in borrowing with unclear Behavioral factors can increase consumers’ purposes, missing loan repayments, placing digital vulnerability to risks in digital credit, and lenders loans as a lower repayment priority, or neglecting or may purposefully or inadvertently exploit these misunderstanding loan terms. vulnerabilities and increase borrowers’ exposure to, and materialization of, risks. Consumers often Default Settings and Status Quo: Consumers often make decisions that deviate from their intentions, accept default conditions without review and may expectations, or best interests, particularly when reborrow out of habit, not need, which perpetuates influenced by the speed, ease, and convenience of dependency or exposure to unfavorable terms. digital credit. These conditions can amplify behavioral biases that are cognitive and emotional shortcuts that Overconfidence and Mental Accounting: Borrowers simplify decisions but often result in irrational or risky may treat different types of debt separately and financial behavior. Research shows that rapid borrowing take multiple small loans (e.g., BNPL products), while decisions, such as those involved in digital credit, remaining overconfident in their repayment ability, heighten susceptibility to such biases (Bartholomae ignoring their cumulative debt burden. and Fox 2021; Hamid 2025; Lea 2020; Livshits 2020; Mazer et al. 2014). Key behavioral biases include: New and Evolving Consumer Risks Hyperbolic Discounting and Present Bias: Consumers in Personal Digital Credit Could Lead to Negative Outcomes often overvalue short-term gains and undervalue long-term benefits and costs, leading to impulsive borrowing. The ease and speed of digital credit can Increased digitization can exacerbate and introduce override self-control and focus on the immediate new and evolving consumer risks in personal digital appeal of having the money, while downplaying or credit. These include fraud, data misuse, unfair postponing concerns about its costs. treatment, lack of transparency, and inadequate redress—often intensified by behavioral vulnerabilities. Anchoring and Message Framing: Consumer choices are Cross-cutting risks such as network downtime, agent- often influenced by how information is presented, such related risks, and other third-party risks pose further as promotional messages that emphasize high maximum challenges. Table 6 below outlines these key risks. loan amounts or ease of access, as they make decisions. These push marketing tactics can lead to consumers Without adequate consumer protection, behavioral borrowing amounts beyond their actual needs. vulnerabilities can be exploited, increasing consumers’ exposure to risks in digital credit. These risks may lead Loss Aversion and Availability Bias: Consumers to overindebtedness and other negative customer often overvalue something readily accessible, with outcomes such as privacy violations, which could unrestricted access. Borrowers may be more driven undermine the financial health of borrowers (see to take out digital loans, simply to avoid missing an Figure 3). 2 opportunity, after reading messages like “you qualify for . . .,” even if the loan isn’t necessary. 2 Overindebtedness is a complex, multi-dimensional concept without a universally accepted definition. A consumer may become over- indebted for many reasons, including a provider’s conduct, unexpected life events, borrower’s financial decisions, and structural poverty (World Bank forthcoming). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 8 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 6. Key consumer risks exacerbated by digital credit Consumer Risk and Issue Example of How Digital Credit Can Exacerbate Consumer Risks Behavioral Vulnerabilities Borrowers’ behavioral vulnerabilities make them more susceptible to complex consumer risks in digital credit (see page 8). Fraud Fraudulent apps or platforms, phishing, spoofing, or fake SMS or websites may trick borrowers into sharing sensitive information (e.g., account numbers, PIN), transferring funds (e.g., loan “application” fees), or allowing access to other apps or device data. Around 40 percent of 1,200 survey respondents in Rwanda reported being targeted by fraudsters. Over 20 percent of banked individuals in Kenya, Nigeria, and Uganda reported financial losses due to fraud, scams, or bribes (CEGA and IPA 2024). Data Misuse Borrowers may have personal data collected or shared without consent or legitimate purpose, utilized for aggressive marketing, sales, or debt collection. Over half of the largest digital loan apps in 2020 collected sensitive data (Kelly and Mirpourian 2021). The average data breach cost increased from USD 4.45 million in 2023 to 4.88 million in 2024, with 46 percent of breaches involving personal data (IBM 2024). Lack of Transparency Hidden, misleading, and noncomparable disclosure of costs affects digital borrowers, who often have limited financial and digital literacy and are subject to high-pressure sales practices. A fifth of surveyed digital borrowers in Kenya and a quarter in Tanzania reported experiencing poor transparency in 2017. Poor transparency correlated with higher late repayment and default rates (Izaguirre, Mazer, and Kaffenberger 2018). Inadequate Redress Borrowers may find it harder to know where and how to complain and obtain redress as more actors participate in offering digital credit. For example, with BNPL products, customers may believe the financing comes from the consumer goods company when it is actually provided through a partnership with a lender, potentially leading to confusion during the grievance/redress process. Unfair Treatment Lenders, credit bureaus, and third parties may discriminate or blacklist debtors, or fail to follow minimum responsible lending practices (Cerise+SPTF 2024; OECD 2019a). Cross-Cutting Risks Agents may conduct irresponsible practices (e.g., overcharging at loan disbursement, aggressive communications with borrowers). Network downtime may affect the borrower experience (e.g., failure to complete a repayment transaction). Other third parties may also conduct or contribute to irresponsible practices (e.g., unethical debt collectors, apps with weak security that enable borrower data leaks). Overindebtedness Greater vulnerabilities and risk exposures may lead to overindebtedness, which can manifest in loan stacking (i.e., multiple borrowing within a short period to repay existing loans), skipping meals, selling assets, or failing to pay children’s school fees (see Figure 3). Sources: Authors, adapted from Chalwe-Mulenga et al. 2022. Note: Consumer risks are adapted from CGAP’s DFS consumer risk typology and include additional challenges related to digital credit: unfair treatment, behavioral vulnerabilities, and overindebtedness. In jurisdictions with more advanced digital credit erode a provider’s loan portfolio quality, it can cause markets, such outcomes are already emerging (see negative spillover effects, where other actors’ solvency Box 2). Overindebtedness not only affects individuals, is affected and providers begin to classify certain it also carries broader implications for financial consumer groups as unlendable (Basel Committee on inclusion and stability. If high delinquency rates Banking Supervision 2016; Tomilova et al. 2018). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 9 Section I Section II Section III Section IV Section V Appendix I Appendix II FIGURE 3. From vulnerabilities to overindebtedness Factors that increase consumer Behavioral vulnerabilities Other vulnerabilities risks when exploited by FSPs Fraud Data misuse • Rogue lenders • Privacy intrusion • PIN sharing • Social shaming Broad risks that Lack of transparency Inadequate redress contribute to • Dark patterns • Uncertainty overindebtedness • Rates (APR) not disclosed • No redress available Unfair treatment • Blacklisting • Irresponsible lending Agent-related risks Repeated calls to borrowers, close relatives and friends Cross-cutting risks that contribute to Network downtime overindebtedness Transaction (repayment) failure when the system is slow or down Other third-party risks Irresponsible debt collection; weak borrower data protection Over-indebtedness Monetary and Financial health Negative outcomes privacy losses deterioration Source: Authors, adapted from Chalwe-Mulenga et al. 2022. BOX 2. Data on overindebtedness in digital credit • From 2014 to 2016, 2.7 million Kenyans defaulted, basic expenses to repay loans (Spriggs and including 400,000 borrowers who had taken out Kipkemboi 2024). digital loans of KES 200 (~USD 2) or less. By the • Approximately 20 percent of digital borrowers end of 2022, 14 million accounts had defaulted on defaulted in Tanzania from 2016 to 2018. digital loan apps (Maringa and Jalloh 2023; Mustafa Late-repaying customers who got another loan et al. 2017). were more likely to repay it late and incur higher • Phone surveys in 2024 showed that 86 percent, penalty fees (Izaguirre, Mazer, and Graham 2018). 70 percent, 55 percent, and 53 percent of digital • Many suicides due to unethical debt collection by borrowers in Kenya, Tanzania, Ghana, and Côte digital loan apps have been reported in India since d’Ivoire, respectively, faced repayment challenges. 2020, including 60 borrowers who committed Defaults ranged from 17–46 percent. About 60 suicide due to receiving abusive messages and percent of Kenyans, 30 percent of Ghanaians and threats (Agarwal et al. 2023). Ivorians, and 20 percent of Tanzanians reduced Responsible Digital Credit: Frontier Solutions for Authorities and Providers 10 Section I Section II Section III Section IV Section V Appendix I Appendix II A Responsible Ecosystem 3. This complex, fragmented framework leads to Approach Is Needed for inconsistent consumer protection standards. Digital borrowers may face heightened exposure to Borrowers to Fully Realize the risks and negative outcomes due to inadequate and Benefits of Digital Credit inconsistent standards. Some digital lenders exploit regulatory arbitrage—choosing their legal form From a consumer protection perspective, three key (e.g., MFI, financing company, fintech) based on developments in digital credit have rendered traditional the most lenient consumer protection regulatory frameworks ineffective to address consumer risks: regime. Rogue actors may also take advantage of 1. Different types of lenders have been operating fragmented oversight to launch fraudulent apps under the remit of different authorities. While that deceive users into sharing personal data or a bank is supervised by the central bank, a sending money, undermining trust in the digital microcredit institution may be overseen by a credit market. sectoral authority, a cooperative by the agriculture ministry, a money lender by a general consumer As a result of these developments, a holistic and protection authority, and a peer-to-peer lender by proactive consumer protection approach to digital the securities regulator. Each authority typically credit is needed. Namely, an RDFE approach to digital applies a different regulatory and supervisory credit, whereby financial and nonfinancial authorities, framework––except where there is an overarching digital credit providers and associations, consumer financial consumer protection law. representatives, and market facilitators (e.g., research organizations) collaborate in: 2. This fragmented framework has worsened with digital credit. Any lender can partner with • Issuing customer-centric standards and rules a fintech to offer digital credit products. New • Monitoring consumer risks local and international fintechs may emerge • Addressing borrower complaints and offer digital credit apps that any customer can easily download from an app store to their • Resolving overindebtedness, privacy, and monetary mobile phones. While these apps may not fall losses. under the remit of a financial authority, they are often subject to regulations issued by data This approach builds on global good practices, protection, cybersecurity, competition, general principles, and standards (e.g., BTCA 2021; Cerise+SPTF consumer protection, financial intelligence, 2024; OECD 2019a; OECD 2022). telecommunications, or ICT authorities. E-commerce platforms and retail stores operating outside the purview of a financial authority may deliver digital credit as an ancillary service to their primary business, while still serving as the main customer interface. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 11 Section I Section II Section III Section IV Section V Appendix I Appendix II SECTION II Consumer Risks in Personal Digital Credit: A Customer-Centric View A S PART OF AN RDFE, IDENTIFYING • Risk Prevention: Implementing measures to reduce solutions to consumer risks in digital credit is or eliminate the likelihood that identified risks will strengthened by a customer-centric view of materialize and negatively affect digital borrowers. the borrower journey. A detailed and holistic view of • Risk Mitigation: Applying strategies to lessen the the customer journey helps stakeholders understand severity or impact of risks that cannot be fully the complexity of digital credit experiences, identify prevented, helping digital borrowers withstand multiple sources of consumer risk, and uncover potential harm. opportunities to develop and adopt solutions. This view • Risk Resolution: Taking action to address risks that entails a better understanding of: have materialized, with the goal of eliminating their • The full consumer journey, from application to negative effects on digital borrowers. repayment and closure, including risks at each stage and potential detours where problems arise and human support is needed. Phases of the consumer risk management FIGURE 4.  • The provider life cycle behind the journey, cycle in digital credit highlighting stages that rely on technology or automation, enable customer feedback, or involve Risk identification partnerships. • The complex regulatory landscape associated with digital credit and various actors that influence different stages throughout the customer journey. Risk Risk In addition, addressing consumer risks in digital credit resolution prevention Risk requires considering four phases of risk management: management • Risk identification (Including Risk Monitoring): Recognizing and measuring potential risks to digital borrowers and tracking them over time to ensure timely detection and response to changes in the risk Risk environment. mitigation Responsible Digital Credit: Frontier Solutions for Authorities and Providers 12 Section I Section II Section III Section IV Section V Appendix I Appendix II Customer Journey in Personal Digital Credit A customer-centric view of personal digital credit signing up for a loan and after a disbursement, such requires a comprehensive understanding of the as researching, seeking support, or resolving issues. different steps a consumer may take to acquire, use, These steps may involve not only the lender but also and close a digital loan. This means going beyond the various partners. Figure 5 illustrates a potential positive typical point-of-sale interaction and recognizing the customer journey, outlining key stages before, during, multiple actions that a consumer may take before and after the digital loan experience. FIGURE 5. Customer journey in personal digital credit: Positive path General awareness Advice / orientation • Sees ads, articles online, collaterals • Checks finfluencer video • Word of mouth • Asks for advice on social media • Asks agents for information Reception of invitation Comparison • Receives SMS, chat, email • Checks marketplace website • Learns of credit option at point • Checks lenders websites of sale Sign-up Application & onboarding Agreement • Downloads app • Indicates key terms of • Agrees to T&C (confirmation of info, • Enters personal data  loan request (amount repayment method, and other T&C) and term) • Chooses disbursal method Timely full repayment Partial repayment Use of funds Approval & disbursal • Fully repays loan • Pays instalment or • Cashes out or • Receives notification of • Can receive offer of partial amount of loan uses digital wallet approval, confirmation, and other products or instructions for disbursal bigger loan Closure of loan Recourse, query & grievance • Receives confirmation of loan closure • See Figure 8 Loan Lifecycle Before loan sign-up During loan After loan closes Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 13 Section I Section II Section III Section IV Section V Appendix I Appendix II Depending on the digital credit business model, a An understanding of the multiple steps that a borrower borrower may still have multiple human, physical may take in their journey can help identify the different interactions with lenders or partners at different types of risks and issues that a customer may face. See stages of their journey. See Figure 6 for an illustration Figure 7 for some examples of these risks and Box 3 for of this path. customer quotes on negative experiences. FIGURE 6. Customer journey: Positive path with human touchpoints General awareness Advice / orientation A potential borrower seeks advice from friends and family via word of mouth. Reception of invitation Comparison Sign-up Application & onboarding Agreement An agent visits the borrower in her store to address questions and help her download the app. The agent can also verify physical KYC documentation. Timely full repayment Partial repayment Use of funds Approval & disbursal The customer visits a nearby agent to cash out, and to cash in for repayment. Debt collection agent may contact late borrowers. Closure of loan Recourse, query & grievance When a customer calls with a complaint, an agent comes by to follow up. Loan Lifecycle Before loan sign-up During loan After loan closes Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 14 Section I Section II Section III Section IV Section V Appendix I Appendix II FIGURE 7. Customer journey: Examples of risks General awareness Advice / orientation Mass marketing to consumers with little Online credit products may be marketed in assessment of individual circumstances or ability a misleading way (e.g., over-emphasizing to repay ("lend to learn" digital credit models). benefits, hiding risks, or giving unrealistic offers with hidden conditions). Reception of invitation Comparison Push marketing leads to customer taking out loan Information on website is outdated or without intentional purpose, due to behavioral does not express the total cost of a biases such as present bias and loss aversion. digital credit in a comparable manner. Sign-up Application & onboarding Agreement Lender collects personal info T&C are hard to understand Timing and flow of from handsets, browser history, and/or hard to access (e.g., links information: key information etc. without customer consent. to T&C provided on a separate such as pricing provided after website). Agent poorly explains completion of transaction. loan terms (interest rates, repayment schedules, penalties). Timely full repayment Partial repayment Use of funds Approval & disbursal Lender (or agent) Transaction (repayment) Customers who Nearly frictionless user harasses a borrower via failure when the network borrowed with limited experience leads to easy excessive calls or is down. Agent calls need or intention have access to credit and may reminders sent to contacts (friends, family) an increased risk of lead to overindebtedness. borrower’s mobile phone. to shame borrowers into non-repayment and full repayment. overindebtedness. Closure of loan Recourse, query & grievance Digital lenders often encourage customers to See Figure 8. roll over loans or to take out larger loans, e.g., by extending the loan automatically when payments are missed, including penalties. Loan Lifecycle Before loan sign-up During loan After loan closes Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 15 Section I Section II Section III Section IV Section V Appendix I Appendix II When digital borrowers face issues during their journey—such as difficulty repaying a loan—they aim to contact the lender for support. Figure 8 illustrates the customer experience and potential risks that a borrower may face at each step of the grievance and redress process. FIGURE 8. Customer journey: Grievance and redress General awareness Advice / orientation Reception of invitation Comparison Sign-up Application & onboarding Agreement Approval & disbursal Timely full repayment Partial repayment Use of funds Closure of loan Recourse, query & grievance In case of borrower problems at any point during customer journey Voice Repayment stress Advice / orientation Late payment redressal • Makes • Does not repay loan • Goes to debt • Faces debt collection inquiry/complaint on time counselling • Asks for debt through • Contacts or responds • Seeks info online restructuring or web/chat/phone/agent to lender rollover Examples of customer risks re: grievance & redress Voice Repayment stress Advice / orientation Late payment redressal Limited knowledge Agents repeatedly call Confusion on who to Lack of recourse about how to complain, and harass borrowers approach. regarding unauthorized who to contact (due to who are late or in activities, such as modularization of DFS default. Overindebted abusive collection and multiple providers customers may skip practices (e.g., shaming involved in service meals, sell assets, or on social media or delivery), and what to fail to pay children's agents contacting do if a complaint is not school fees. relatives and friends.). solved satisfactorily. Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 16 Section I Section II Section III Section IV Section V Appendix I Appendix II Provider Life Cycle in Digital Credit Customer Quotes on Negative Experiences BOX 3.  A clear understanding of the customer journey in with Digital Credit digital credit is closely tied to the provider’s life cycle. “@zeebusiness @AnilSinghvi_ #OperationHaftaVasooli Each stage—from onboarding to repayment—relies Sir kisi door ke relative ne credit card ka bill nahi pay on real-time or near-real-time interventions, with risk kiya They found me on Facebook and started calling checkpoints embedded throughout the life cycle. me on my mobile also on landline and using very bad Data-driven feedback loops further enhance this process, abusive language. I don't know what to do.” continuously informing and refining product design, credit “This number 8792662456 from my cash app. Call assessment, and customer engagement strategies. me back to back and call my contacts without my permission. They calling my contacts and misusing Digital credit providers increasingly rely on a blend of my personal data. #BanLoanApps @RBI @RBIsays alternative and traditional data sources –such as mobile @CyberGujarat @DelhiPolice @GujaratPolice @ phone metadata, transactional behavior, and formal BanegaAb.” credit histories—augmented by AI, ML, and application “I didn't go to the terms and conditions because you programming interface (API) technologies to streamline have to be connected to the Internet. And well, do the credit life cycle. These technology integrations you have time to look at all that in the circumstances enhance the delivery of seamless customer experiences under which you apply for a loan?” (Anonymous while proactively detecting and mitigating risks across male, self-employed). the credit life cycle (see Figure 10). To operationalize Sources: Duflos et al. 2023; Navarro et al. 2024. these capabilities, digital credit providers often partner with third-party vendors. While these partnerships can enhance scale and efficiency, they also introduce critical considerations around data security, operational dependency, and regulatory compliance that providers must actively manage. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 17 Section I Section II Section III Section IV Section V Appendix I Appendix II FIGURE 9. Provider life cycle stages involved in delivering digital credit Product design and testing Marketing and outreach Data-driven product design Enhanced customer outreach & literacy activities based on data analytics Onboarding Verification, KYC Credit assessment Pricing Repayment behavior and reporting feedback into risks assessment models Fraud monitoring (flagging anomalies after disbursement) Fraud monitoring (flagging Fraud prevention (blocking anomalies after disbursement) suspicious loan applications) Repayment Disbursement Approval and offer Terms disclosure Repayment behavior can inform credit assessment model, update credit bureaus and reporting agencies, and lead to debt collection processes (e.g., reminders, restructuring options, legal proceedings, external recovery) Recourse, query & grievance Recourse can improve product design and terms disclosure. Queries can generate insights for customer outreach, debt collection, repayment and recourse mechanisms Loan Lifecycle Before loan sign-up During loan After loan closes Title for these elements? Data-driven Real-time fraud risk Real-time process feedback loop checkpoint Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 18 Section I Section II Section III Section IV Section V Appendix I Appendix II FIGURE 10. Provider life cycle stages and technology touchpoints to address consumer risks in digital credit Product design and testing Marketing and outreach Use of data analytics Chatbots to handle FAQs for acquiring customers and pre-qualification checks Onboarding Verification, KYC Credit assessment Pricing Biometrics & adaptive Integration with Support for Behavioral analytics, Risk-based pricing authentication, MFA credit bureaus document AI-driven analytics using alternative and national ID uploads and credit scoring, AI systems status updates fair pricing and explainability Repayment Disbursement Approval Terms disclosure and offer Repayment Automated ML and data analytics to Behavioral monitoring repayment preempt late biometrics, AI-driven with real-time reminders and repayments or defaults anomaly detection analytics support chatbots and post-disbursement fraud risk Recourse, query & grievance First-line customer support for complaints, queries, and grievances Loan Lifecycle Before loan sign-up During loan After loan closes Title for these elements? Technology integration: AI, ML, AI-driven Chatbots APIs, Data analytics Source: Authors, building on Stewart et al. 2018; D91 Labs N.d.; CGAP 2025b. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 19 Section I Section II Section III Section IV Section V Appendix I Appendix II The Fragmented Regulatory and Furthermore, several authorities may have different Institutional Arrangements in responsibilities for various actors in the digital credit ecosystem at different stages of the provider’s life Digital Credit cycle. Authorities can set, enforce, and monitor compliance with responsible digital credit standards In a typical jurisdiction, a digital borrower may be that shape consumer experiences, risks, and outcomes. directly or indirectly engaging with a range of credit Table 7 offers an example of the different types of services providers. Different authorities would have authorities that may be involved in regulating or regulatory remits over all these entities, with disparate supervising different types of actors involved at consumer protection rules and supervisory activities, different stages of the digital credit provider life cycle. which leave borrowers (especially vulnerable, unserved, and underserved segments) exposed to greater consumer risks that may lead to adverse outcomes. Figure 11 illustrates the different types of authorities (financial, telco/ICT, and general) that may fully or partially regulate and supervise different types of providers offering digital credit, leading to overlaps and gaps in institutional mandates. FIGURE 11. Fragmented regulatory and institutional arrangements in digital credit Telco/ICT authority Financial sector authority MNOs delivering loans General consumer protection, Platforms lending competition and data money (e.g., protection authorities e-commerce, ride-hailing) FSPs originating Fintechs providing Unregulated money digital loans platform / ancillary lending apps services Responsible Digital Credit: Frontier Solutions for Authorities and Providers 20 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 7. Regulatory and institutional arrangements and provider life cycle in digital credit Providers Product Marketing Onboarding Credit Terms Disbursement Repayment Recourse, design and and assessment, disclosure and credit query, and and outreach verification pricing, reporting grievance testing approval, and offer Digital credit Financial FSA FSA and FSA FSA FSA FSA and FSA and providers* sector financial consumer consumer authority intelligence protection protection (FSA) unit (FIU) authority authority MNOs Telco and Telco Telco Telco FSA, consumer and data authority and authority consumer protection protection FSA and FSA protection, authorities authorities, and telco FIU authorities Credit FSA and data FSA and bureaus protection data authority protection authority Supply chain Consumer Consumer Consumer FSA and FSA and Consumer distributors* protection protection protection consumer consumer protection authority and data authority protection protection authority protection authority authority authorities Debt FSA and FSA and collection consumer consumer companies protection protection authority authority Alternative Consumer Consumer Consumer data and protection, protection, protection, credit score ICT, and data ICT, and data ICT, and providers protection protection data authorities authorities protection authorities Technology Consumer Consumer ICT and data ICT and data Consumer infrastructure, protection protection, protection protection protection, cloud authority ICT and data authorities, authorities ICT, and services protection FIU data authorities protection authorities *May include fintechs, bigtechs, and other platforms. Source: Authors. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 21 Section I Section II Section III Section IV Section V Appendix I Appendix II From a financial regulatory perspective, a digital lender • Illegal: An entity offering digital loans in may fit different categories depending on the applicable contravention of the law (i.e., not falling within one regulatory perimeter. Thus, a digital lender may be: of the above categories), or because its lending business does not conform to regulation. An illegal • Regulated: A lender covered by financial regulations lending app may engage in criminal activities (e.g., or laws that include digital credit as a category loan sharking, fraudulent scams), or may not have any of financial services, either explicitly or implicitly. business in place and use the app only as a facade. The lender is licensed, registered, or approved by the banking regulator or another official agency All these perimeters may coexist in a jurisdiction as part responsible for regulating nonbank lenders. of the evolving landscape. This may lead to different • Unregulated: A firm offering digital credit to borrower experiences, risks, and outcomes depending customers for its goods or services, beyond the reach on the lender they engage with, as well as the different of financial regulation but subject to other broader, standards for different providers. It’s important for general laws (e.g., consumer or commercial codes). authorities to not only assess the implications of these • Informal: An entity with or without legal status different perimeters but also to distinguish them offering digital loans that fall outside the scope of when communicating about digital lenders to avoid law and regulation. This may be due to a regulatory misunderstandings and misperceptions (e.g., conflating gap, a transition phase to become a regulated criminals with informal or unregulated lenders). It’s also lender, or a carve-out for the digitization of important to differentiate between lenders and third traditional credit groups (e.g., ROSCAs, VSLAs). parties, as well as the roles different authorities play in relation to these actors. FIGURE 12. Financial regulatory perimeter of digital lenders Illegal Informal Financial regulatory perimeter Unregulated Regulated Responsible Digital Credit: Frontier Solutions for Authorities and Providers 22 Section I Section II Section III Section IV Section V Appendix I Appendix II SECTION III Frontier Solutions For Responsible Personal Digital Credit: Authorities G OVERNMENT AUTHORITIES and coordination with telecommunications, ICT, and (including financial and nonfinancial sector data protection agencies is vital. This includes regulatory authorities) have a critical role to play in consistency across agencies. addressing the key consumer risks arising in digital credit provision, i.e., consumer protection risks. Digital An effective approach to consumer risks comprehensively loans may be covered by bespoke rules, but are more addresses all phases of risk management: often regulated as consumer credit, money lending, payday loans, or microcredit. Enforcement of these Risk Identification: A supervisory role over rules may fall under different actors, namely: multiple types of FSPs, including the power to monitor the entire digital credit market • A specialized agency (e.g., the United States’ and request information from individual regulated Consumer Financial Protection Bureau [CFPB]), entities to assess consumer risks and outcomes. • A unit within the central bank or a unified financial regulator (e.g., the Australian Securities and Risk Prevention: A regulatory role, setting Investments Commission [ASIC]), and enforcing rules to empower consumers • The conduct regulator in a twin peaks system (e.g., throughout their credit journey and avert the United Kingdom’s Financial Conduct Authority consumer harm, addressing issues such as overly [FCA]), or aggressive and misleading marketing, unsuitable products, unfair terms, bias, nontransparency, • A general consumer protection authority (e.g., overindebtedness, data misuse, and abusive loan Nigeria’s Federal Competition and Consumer collection. Protection Commission [FCCPC]). Risk Mitigation: Providing means to contain In most of the jurisdictions reviewed, the prudential risks incurred, e.g., through adequate dispute authority is not involved unless digital loan risk is held resolution schemes, treatment of reckless or on a bank’s balance sheet. predatory lending, and compensation to fraud victims. Authorities address consumer risks through various Risk Resolution: Providing a pathway to informal and formal interventions, including industry eliminate the impact of negative outcomes collaborations and commissioning or delegating work to on consumers, e.g., through mechanisms to other ecosystem actors, such as consumer associations address personal insolvencies. and research organizations (see Section V). Collaboration Responsible Digital Credit: Frontier Solutions for Authorities and Providers 23 Section I Section II Section III Section IV Section V Appendix I Appendix II Solutions led by authorities include some that cut redress. Other solutions provide complementary across all types of risks and issues, such as ecosystem protections, such as fraud compensation and collaboration, entry controls into the regulated market consumer safeguards in the event of provider (i.e., registration and licensing), and market monitoring insolvency (see Table 8). tools. Most solutions focus on regulations addressing key financial consumer protection elements, including disclosure, product governance, credit scoring and reporting, prevention of lender abuses, and complaint TABLE 8. Authority solutions by phase of risk management cycle and consumer risk Authority Solution Risk Management Cycle Consumer Risk and Issue Identification Prevention Mitigation Resolution 1. Ecosystem collaboration 2. Registration All risks and issues 3. Licensing Adequate disclosure before, 4.  during, and after sales of digital credit Lack of transparency Credit information: scoring 5.  and reporting 6. Positive friction Behavioral vulnerabilities 7. Product governance Unfair treatment 8. Data protection Data misuse 9. Combating fraud Fraud Digital credit market 10.  All risks and issues monitoring 11. Fraud victim compensation Fraud Effective redress 12.  Inadequate redress mechanisms  revention of predatory 13. P lending and abusive debt collection Overindebtedness 14. Personal insolvency A Call to Action for Supervisors, Regulators, and Policy Makers A Call to Action for Funders Responsible Digital Credit: Frontier Solutions for Authorities and Providers 24 Section I Section II Section III Section IV Section V Appendix I Appendix II Responsible Digital Credit Solutions by Authorities Solution 1: Ecosystem Collaboration All risk management phases • All consumer risks and Issues WHY? (THE PA IN P O IN TS ) within the expanded perimeter but do not comply The digital credit landscape is increasingly complex, with registration and licensing requirements may be with diverse lenders partnering with third parties sanctioned or closed. at different stages of the borrower journey. This • Indirect regulation (e.g., through rules on complexity extends to regulation, where multiple outsourcing by a regulated institution, see example financial and nonfinancial authorities operate in silos. from India on page 29). When a regulated FSP For example, central banks supervise banks, consumer delivers digital loans through a partnership with protection agencies monitor fintech marketplaces, an unregulated (nonfinancial) business, such as and telco regulators oversee MNOs delivering loans. an e-commerce platform, the loans are subject to Meanwhile, competition and data protection authorities the financial authority’s outsourcing regulations. assess broader market practices. Consumers face This action will benefit from collaboration with risks from fraudulent apps, overindebtedness due to general authorities responsible for supervising multiple loans, and confusion over where to report such nonfinancial businesses, enabling improved complaints. knowledge sharing, coordinated supervision, and stronger enforcement. WHAT ? (THE SO LUT IO N S ) • Coordination and collaboration among regulatory As indicated on page 20, digital lenders may operate agencies (e.g., see Nigeria and the Philippines outside the regulatory perimeter of financial below). This includes signing an inter-agency authorities. Collaborative approaches can be taken to memorandum of understanding or forming a expand such a perimeter and bring more—if not all— multistakeholder task force, working group, or digital lenders under the remit of financial authorities: informal collaboration forum. This should include nonfinancial authorities with specialized mandates • Expansion of the perimeter to bring digital lenders across economic sectors (e.g., consumer protection, under direct regulation (e.g., see example from telco/ICT, or data protection authorities) that can Malaysia below). This entails establishing an activity- share key insights on market developments and based framework covering all providers of digital emerging consumer risks in digital credit. Any credit, including banks, nonbank credit institutions, collaboration or information sharing should align fintechs and digital lending apps. This framework with the statutory mandates and legal discretion of should set clear financial consumer protection the participating agencies. and market conduct rules tailored to the different types of providers. It should also assign mandates Additionally, there are complementary, nonregulatory to one or more authorities, which may include measures that could be taken –that are discussed general, nonfinancial regulators taking on financial further below (see Sections IV and V). These include: consumer protection responsibilities. Coordination among these authorities is essential to minimize regulatory overlaps and gaps. Lenders that fall Responsible Digital Credit: Frontier Solutions for Authorities and Providers 25 Section I Section II Section III Section IV Section V Appendix I Appendix II • Industry codes of conduct with self-regulatory Ecosystem collaboration helps ensure seamless monitoring and enforcement mechanisms led by coverage of consumer risks in digital credit. At the industry associations. government level, this requires consistent collaboration among financial sector, telecommunications, ICT, • Work with associations of fintechs or digital and consumer protection agencies. These authorities lenders to frame guidance in key areas for digital can work together to develop, implement, and app-based lenders. monitor solutions that reduce gaps, overlaps, and • Consultations with consumer associations to gather inconsistencies. For such coordination to be effective, insights on consumer risks and inputs for guidance, a clear lead authority, designated by legislation or and collaboration on consumer communications. formal agreement, should guide the process, supported by binding commitments of the other actors. These solutions emphasize coordinated action or “ecosystem collaboration,” in which authorities, providers, and other key stakeholders work together C H AL L E N GE S AN D L I MI TAT I O N S to develop, implement, and monitor solutions that Collaboration can be affected by low levels of trust, are more impactful than isolated efforts. Effective negative past experiences, misaligned goals, unclear collaboration reduces regulatory gaps and overlaps mandates, weak communication, and inadequate and leads to better outcomes for consumers. coordination structures. Identifying a committed To be meaningful, such collaboration must go champion, such as the financial sector authority, and beyond symbolic gestures. It requires sustained securing clear commitments from all stakeholders communication, shared responsibilities, and tangible can help address these challenges and strengthen joint actions. collaborative efforts. Examples Malaysia’s Consumer Credit Oversight Board Task Force was launched in 2021 by the Ministry of Finance, Bank Negara Malaysia, and Securities Commission, in collaboration with the credit cooperative regulator and several ministries, to develop a regulatory and supervisory framework and centralized redress mechanism for nonbank credit providers and credit service providers (e.g., BNPL providers, debt collectors) (Consumer Credit Oversight Board Task Force n.d.). In Nigeria, in 2022, the FCCPC together with the Independent Corrupt Practices Commission, the National Information Technology Development Agency, and the Central Bank, formed the Joint Regulatory and Enforcement Task Force to rein in digital credit apps, including through regulatory guidelines (FCCPC 2022a) (see Appendix I). In the Philippines, the central bank, deposit insurer, securities regulator, and insurance commission formed the Financial Sector Forum in 2004 (Funa 2017). The Forum has issued consumer protection guidelines for fintechs, including digital lenders, and worked with industry associations on customer-centricity campaigns. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 26 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 2: Registration of Digital Lenders Risk prevention • All consumer risks and issues WHY? (THE PA IN P O IN TS ) 2. Periodic operational information (e.g., financial Registration is a foundational step in applying statements, scale and portfolio quality indicators) regulations to providers. It certifies that a credit-only and details on the characteristics, terms, and costs FSP is permitted to lend (as its primary business) in of products offered. a situation where it otherwise might not be allowed, or the legal status is unclear. Registration helps Providers must also comply with consumer credit clarify the legal status of lenders, provides assurance regulations, including rules on loan terms, transparency to consumers, and applies minimum standards to and disclosures, fair collection practices, credit lenders’ operations, ownership, and management. It reporting, and complaints handling. Registering also enables authorities to exert basic oversight, such authorities should maintain publicly accessible as control over entry, monitoring, and light-touch databases to allow consumers to verify whether a supervision, including tracking, reporting, spot checks, digital lender is registered and in good standing. and complaint analysis. C H AL L E N GE S AN D L I MI TAT I O N S WHAT ? (THE SO LUT IO N S ) Registration can be effective in environments where Establish a legal or regulatory requirement for providers strong consumer protection laws or regulation and to register with a specific authority to offer digital enforcement already exist. While less burdensome credit as a principal business activity. Registration than licensing, registration also provides more limited requirements should include: oversight. It affords the regulator less ex ante control and enforcement power over the FSP, which may make 1. Basic institutional information (e.g., location, legal it less suitable where digital lenders are scaling up fast status, capital structure, officers, and board) Examples Nigeria’s Interim Guidelines for Digital Lending (2022): In response to abuses by digital lenders, many of them unregulated, Nigeria formed a multi-agency task force to develop regulatory guidelines. The consumer protection agency issued the guidelines, which include a registration requirement. Digital lenders must: Provide basic information on the identity of stakeholders, accounts, funding sources, affiliates, •  lending methodology, data audit, and list of apps (max. 5). • Report any changes to the regulator (FCCPC 2022b). South Africa’s National Credit Regulations (2006): The National Credit Regulator and its rules were adopted in response to rampant loansharking that carried over from the apartheid period. The registration process requires lenders to submit information including ID, tax number, products offered, loan book value and numbers, business premises, staffing plan, administrative procedures, and business records and criminal records (Republic of South Africa 2015). This registration process has helped the National Credit Regulator monitor all digital lenders. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 27 Section I Section II Section III Section IV Section V Appendix I Appendix II or need closer monitoring due to a history of consumer licensing may become necessary. Industry associations abuses and defaults. Ideally, supervisors should monitor can also support the registration process and registered FSPs to identify when certain players, contribute to self-regulation, including enforcing codes subsectors, products, or delivery channels evolve to of conduct (see Section IV, Solution 12). a point where the risks they pose are material, and Solution 3: Licensing of Digital Lenders Risk prevention • All consumer risks and Issues WHY? (THE PA IN P O IN TS ) on-site inspections. With licensing, the government Licensing is typically reserved for prudentially regulated takes on a fiduciary duty and signals the public that institutions. However, the rapid growth of the digital it will effectively supervise the FSP. In digital credit credit industry, along with its expanding client base and markets, this typically means applying existing licensing subsequent increases in consumer risks, may justify regimes (e.g., for credit institutions in general) rather licensing requirements for digital lenders in some than creating bespoke licenses for digital lenders. contexts. Licensing can help safeguard consumers against widespread reckless lending, defaults, and overindebtedness, particularly among vulnerable C H AL L E N GE S AN D L I MI TAT I O N S Licensing may not be necessary for all credit-only consumers, while also helping to mitigate contagion providers, especially where there is no prudential risks to the broader financial sector. risk, i.e., they do not take funds from the public. In such cases, consumer protection codes and ex post WHAT ? (THE SO LUT IO N S ) enforcement may be sufficient. Licensing may also be Licensing imposes more stringent requirements than burdensome for both lenders and regulators, especially registration, including capital, business plans, fit and for small credit-only providers, and in markets where proper assessments, internal controls, operational the sector is at an early stage of maturity. policies, and IT systems. These requirements are more costly and time consuming but serve to restrict market entry to qualified providers, helping to conserve supervisory resources. Licensing also allows for more intensive oversight, including audits, reporting, and Responsible Digital Credit: Frontier Solutions for Authorities and Providers 28 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples Australia’s National Consumer Credit Protection Act (2009) and Regulations (2010): These rules create a unified personal credit regime with stringent requirements for consumer lenders, including mandatory licensing of virtually all consumer credit providers (Australasian Legal Information Institute [AustLII] 2010). License applicants must provide key documentation on, e.g., fit & proper criteria, adequate resources, staff training and oversight, compensation plan, risk management, and dispute resolution. ASIC, the securities regulator, enforces the rules, monitoring compliance with periodic checks. India’s Guidelines on Digital Lending (2022): The Reserve Bank of India (RBI) issued these guidelines as a response to digital lender abuses and fake apps. The guidelines define digital credit, regulate digital lending by all FSPs licensed by the RBI, and apply “indirect” regulation to third- party intermediaries and actors handling digital credit activities outsourced by licensed FSPs (RBI 2022a). Digital lenders must have a Financial Institution license, but there is no bespoke licensing for digital lending. Kenya’s Digital Credit Providers Regulations (2022): The rise of digital credit in Kenya during the 2010s led to rampant abuses, defaults, and overindebtedness. Initial emergency measures led to the adoption of regulatory guidelines that require licensing of digital lenders not already licensed as banks or other FSP types and set standards for governance, lending practices, consumer protection, and credit information sharing (Central Bank of Kenya [CBK] 2022). The Business Laws (Amendment) Act (2024) extended the Central Bank’s authority and the credit licensing regime to all non-deposit taking credit providers (Republic of Kenya 2024). Solution 4: Adequate Disclosure Before, During, and After Sales of Digital Credit Risk prevention • Lack of transparency WHY? (THE PA IN P O IN TS ) to deceptive practices. Research shows that lack In a highly competitive digital lending environment, of transparency is linked to lower repayment rates providers may prioritize rapid growth and short-term (Izaguirre and Mazer 2018). While civic pressure and profits by engaging in marketing and sales practices codes of conduct may partly address the problem, that overstate loan benefits and obscure credit public authorities play a critical role in setting costs and risks. As a result, information presented minimum standards, preventing abuse, and ensuring in advertisements, notices, point-of-sale, and information is consistent, comparable, and accessible other disclosures is often incomplete, inaccurate across the sector. While disclosure is a concern or misleading, or difficult to understand, especially in every kind of lending, the speed of approval, on mobile devices. This disproportionately affects automation, and reliance on electronic interfaces pose underserved consumers, who may face lower financial greater challenges. and digital literacy levels and are more vulnerable Responsible Digital Credit: Frontier Solutions for Authorities and Providers 29 Section I Section II Section III Section IV Section V Appendix I Appendix II WHAT ? (THE SO LUT IO N S ) and easy to navigate, and “dark patterns” (user Regulatory frameworks on disclosure aim to ensure that interfaces designed to deceive or nudge users into consumers clearly understand key digital credit terms making unintended, potentially harmful choices) are and conditions (T&C) throughout the borrower journey. prohibited. Lenders are encouraged to use layering Effective disclosure rules typically address the following: to facilitate understanding, and use behavioral insights to encourage consumers to engage with 1. Content: Rules require prominent disclosure of T&C, information. Lenders must explain the T&C orally with total cost metrics, a clear breakdown of costs upon request or when deemed necessary based on (such as finance charges and third-party fees), and a customer’s circumstances.3 provisions on changes and penalties. The language must be simple and comprehensible to all consumers. 2. Timing: Rules require full disclosure before a loan C H AL L E N GE S AN D L I MI TAT I O N S is contracted, while prohibiting deceptive ads Mobile interfaces often limit consumers’ ability to and notices. The periods during and after the access and understand full T&C. Key information may agreement is made are also covered, ensuring that be buried on different pages, screens, or links—which T&C remain consistent, and any changes are clearly may not always be easy to find or navigate. Complex noted and explained. presentation formats and manipulative design features, such as “dark patterns” can confuse consumers or lead 3. Format: Rules mandate plain language and them to unsuitable products. Low digital and financial standardized presentation, while requiring disclosure literacy means that improving disclosures may have of key T&C in the same channel used for the limited impact unless paired with consumer education. transaction. User interfaces must be user-friendly Examples Content: In the EU, consumer lenders must disclose all key information—including credit terms, rights, duties, penalties, and bundled services with opt-ins or opt-outs—on a standard form prior to entering an agreement. Consumers have a right to receive adequate explanations and human intervention before agreeing (European Union 2023a).  Format: The EU prohibits deceptive “dark patterns” in financial contracts concluded at a distance (where they have been most prevalent). Electronic display of precontract information must be designed in a way that avoids confusion and should use “layering” to assist understanding (European Union 2023b). Content: The Hong Kong Monetary Authority (HKMA) issued a circular in 2022 strengthening guidelines on BNPL products, including on disclosure, chargebacks, and creditworthiness assessments. In marketing a BNPL product, lenders must disclose that it is a “credit product” (HKMA 2022). Format: HKMA’s circular on consumer protection in digital marketing activities requires FSPs to ensure that all advertising and promotional materials are fair and reasonable and do not contain misleading information. 3 Where key information is provided upfront with cross-references or links to further detail. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 30 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples (continued) Content: The United Kingdom’s Financial Conduct Authority’s “consumer understanding” outcome requires providers to ensure that their communications meet the customers’ information needs, i.e., are likely to be understood and equip them to make decisions that are effective, timely, and properly informed (FCA 2022). Thus, all FSPs must test, monitor, and adapt communications to support understanding and optimal outcomes for customers. Timing: Kenya’s Competition Authority found that DFS providers were not disclosing the cost of transactions until after the consumer accepted the transaction via their mobile device (Nkhonjera 2017). The authority issued guidelines requiring all DFS providers to disclose all applicable charges to customers for any mobile money transaction (including microloans, money transfers, and microinsurance) prior to the transaction’s completion (see Appendix I). Format: In Portugal, digital lenders must prominently present information on the basic product features and other elements deemed relevant, such as fees and expenses, on the main screen or webpage of the marketing platform, using larger characters, information boxes, pop-ups, cost simulations, overviews, or other similar means. Providers also must assist customers to obtain further information by making available tools such as a hotline or live chat, chatbot, or other interactive tools (Banco de Portugal 2020). Solution 5: Credit Information: Scoring and Reporting Risk prevention • Lack of transparency WHY? (THE PA IN P O IN TS ) both consumers and regulators to assess algorithmic Many unserved and underserved consumers in systems (Fernandez Vidal and Barbon 2019; Gambacorta developing countries lack formal credit histories, et al. 2019). documented income, or access to traditional bank accounts. Standard credit reporting systems often exclude retail purchases on credit, utilities and rent W H AT ? ( T HE S O L U T I O N S ) Regulation of algorithmic credit scoring may: payments, and small loans, thereby excluding a significant portion of the population from formal credit access. To fill • Apply fair treatment, transparency, and this gap, lenders and credit scoring providers increasingly antidiscrimination standards to algorithms (e.g., US, rely on algorithmic credit assessments using alternative see below). data, such as mobile money transactions and balances, • Require appropriate safeguards during development, mobile phone activity, and social media data. While these testing, and deployment of algorithms to assess models, often powered by AI/ML, can improve credit risk and manage consumer risks (e.g., Hong Kong, see prediction when trained on robust datasets, they also below). introduce new risks. These include biased outcomes due • Require regular auditing of algorithmic systems by to flawed design or data inputs; unfair discrimination external experts (e.g., EU, see below). based on gender, race, religion, etc.; and the inability of Responsible Digital Credit: Frontier Solutions for Authorities and Providers 31 Section I Section II Section III Section IV Section V Appendix I Appendix II • Give consumers the right not to be subject solely credit information, giving consumers a clean slate to automatic processing and the right to request after several years—important for small borrowers human intervention (e.g., EU, see below). blacklisted due to past negative credit history (e.g., South Africa) (Campbell Attorneys n.d.). On the data input side, regulators have mandated (or strongly encouraged) comprehensive credit reporting. This means: C H AL L E N GE S AN D L I MI TAT I O N S Ensuring transparency in algorithmic credit scoring • Ensuring that all creditors report on credit balances and decision–making is challenging, especially when (e.g., Australia, see below). models rely on “black-box” AI systems. In principle, • Reducing or eliminating minimum threshold regulators should assess the governance process, amounts for credit reporting (e.g., Bangladesh) including design and implementation of algorithms (World Bank Group 2023). used in lending, or at least their performance. However, regulators typically consider this task too complicated • Full-file reporting, i.e., requiring all creditors, not and technical and avoid taking action. only FSPs, to report both positive and negative information on borrowers (e.g., Australia, see below). Input data from credit reports can be made fairer to small borrowers by regulations mandating comprehensive Some regulators have shielded vulnerable first-time full-file reporting. But where this requirement brings borrowers and those experiencing hardship by setting in data on small borrowers, it may include defaults on thresholds designed to keep de minimis amounts out “lend-to-learn” digital microloans, resulting in years-long of the system or by restricting information on items unfair blacklisting of vulnerable customers. such as delinquent medical bills (e.g., Kenya, see below). Some have also set maximum retention periods for Examples Credit Assessment and Scoring: One of HKMA’s consumer protection principles for the use of big data analytics and AI establishes that models shall produce fair outcomes using appropriate weights for all relevant variables and allow manual intervention to mitigate irresponsible lending decisions (HKMA 2024). Credit Assessment and Scoring: The US Consumer Financial Protection Bureau (CFPB) enforces federal antidiscrimination law applicable to financial services (CFPB 2022a). The law requires companies to explain to applicants the specific and accurate reasons for denying a credit application or taking any other adverse actions, even if they were made using complex “black-box” algorithms. Creditors cannot use technologies in decision-making if they are unable to provide the required explanations (see Appendix I). Credit Assessment and Scoring: In India, licensed FSPs (banks and nonbanking financial companies) must capture the economic profile of the borrowers (covering age, occupation, income, etc.), before extending any loan, with a view to assessing the borrower’s creditworthiness in an auditable way (RBI 2022a). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 32 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples (continued) Credit Assessment and Scoring: The EU’s Artificial Intelligence Act (2024) provides risk-based AI regulation based on its potential to cause harm. It includes financial use cases, in particular creditworthiness assessments. Applications are categorized as low-risk, high-risk, and illegal (European Parliament 2024). The EU’s General Data Protection Regulation (GDPR) stipulates the rights of consumers not to be subject to a decision based solely on automated processing, and to obtain human intervention, express their viewpoint, and contest the decision (European Union 2016). The European Central Bank’s fintech credit licensing regime includes an assessment of the lender’s capacity to audit outsourced credit scoring activities (European Central Bank 2018). Credit Reporting: The Central Bank of Kenya’s digital credit regulations provide that no negative credit information may be submitted to a credit bureau for any loan where the outstanding amount due is KES 1,000 (USD 7.50) or less (CBK 2022). This protects borrowers who received digital loans from “lend-to-learn” models, so that they would not be blacklisted for small delinquencies. Credit Reporting: Australia’s law on Comprehensive Credit Reporting requires all lenders to provide full-file reporting on consumer credit accounts, and enables credit bureaus to collect information on customers from financial institutions and telephone companies (World Bank Group 2023). Solution 6: Positive Friction Risk prevention • Behavioral vulnerabilities WHY? (THE PA IN P O IN TS ) designed to “intentionally slow users down” and help Automation increases speed and certainty in digital shift them from automatic thinking to more conscious credit but eliminates key human interactions that choices that support positive consumer outcomes can prevent consumers making poor credit choices. (Venkatesan et al. 2024). Inappropriate, unsuitable credit products may be pushed onto consumers through apps, SMS, or aggressive marketing without adequate explanations of features W H AT ? ( T HE S O L U T I O N S ) Policy makers can promote positive friction through and costs of credit. Without human engagement, FSPs targeted consumer protection measures that include: may miss important contextual information about a borrower’s needs, constraints, and capability. This • Pricing transparency and disclosure rules that increases the likelihood that consumers will choose provide an extra review of T&C or comprehension credit products that do not meet their needs. tests to improve consumer understanding of their loan obligations (see UK example). Interposing delays and extra steps to improve digital • Cooling-off and withdrawal provisions that help credit outcomes is referred to as “positive friction” consumers reflect on their product choices and gather (Venkatesan et al. 2024). These friction points are more information (see examples from EU and India). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 33 Section I Section II Section III Section IV Section V Appendix I Appendix II • Increased protections for vulnerable consumers, C H AL L E N GE S AN D L I MI TAT I O N S such as preventing the use of a digital credit Lenders find it more efficient and profitable to rely product during periods of susceptibility or for wholly on automation. This results in small, short-term, undesired behaviors (e.g., using digital loans for high-cost loans with rapid approval and rollover that gambling—see example from Latvia). may become delinquent. In the near term, positive friction imposes costs that providers and consumers would rather avoid, but in the longer term, it promotes a safer and more stable digital credit market with lower rates (and costs) of default. Examples EU: In the EU, consumer credit directives require the right of withdrawal from distance contracts to be made available to the customer through a prominent, easy-to-find “withdrawal function” on the FSP’s interface. It must be comprehensible, clearly labeled, and continuously available during the withdrawal period (European Union 2023b). If a borrower withdraws, they must receive a confirmation receipt without delay (see Appendix I). India: In India, during the cooling-off/look-up period, the borrower must be given an explicit option to exit the digital loan by paying the principal and proportionate APR without any penalty (RBI 2022a). Cooling-off periods are at least three days for loans repayable in seven days or more, and one day for loans repayable in fewer than seven days. Latvia: The Ministry of Economy’s 2015 amendments to the Latvian Consumer Rights Protection Law prohibit online consumer lenders from granting loans from 11 p.m. to 7 a.m. and require them to conduct a creditworthiness assessment based on sufficient information provided by the consumer (Likumi 1999). UK: FCA experimented with positive friction techniques in investments, e.g., by adding checkboxes and questions to confirm customers’ understanding of investment risks. While users improved their product understanding, some responded by dropping out and not completing the process (Venkatesan et al. 2024). The experiment informed FCA’s Consumer Duty Guidelines, which require FSPs to prioritize consumer needs (FCA 2022). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 34 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 7: Product Governance Risk prevention • Unfair treatment WHY? (THE PA IN P O IN TS ) • Assess customers carefully and only offer credit Product governance refers to a lender’s systems, products that are affordable, fit for purpose, and procedures, and controls in place to design, approve, adequately understood; and distribute, and assess financial products. In digital • Have controls in place to verify whether credit credit, product governance frameworks aim to products or services are operating as intended, and ensure that loans are suitable, i.e., they meet the if not, to take action accordingly. actual needs, objectives, and constraints of the consumer. These frameworks help embed good By holding lenders accountable for the suitability customer outcomes into the FSP’s incentives, rules, of their products, this approach avoids the need for and practices, to deter borrower harm and ensure that regulators to prescribe specific terms for each type borrowers are treated fairly. of digital credit product. Guidance may be issued to clarify or remind lenders that they must have effective processes for design, testing, review (including WHAT (THE SO LU T IO N S ) customer feedback), and approval of digital credit Product governance standards place responsibility on products and services before they are offered to the the digital lender to: public. Product approval may be given by the lender’s • Understand the consumers in its target market and board or the authority. design digital credit products and services that meet the customers’ needs; Examples Australia: Before entering a contract, a credit provider must assess the suitability of the credit contract or consumer lease and, if requested by the consumer, provide the assessment in writing (Australian Government 2024). The transaction may proceed only if it is “not unsuitable” for the consumer. India: RBI’s Digital Lending Guidelines address broader market incentives for suitable digital lending (RBI 2022a). They include a prohibition on most forms of loan securitization, which helps preserve lenders’ stake in the performance of their loan portfolios (though it may constrain their access to cheap loanable funds) (RBI 2022b). Singapore: Rules on unsecured credit to individuals took a “bright line” approach in 2013, imposing ceilings on lending and debt service based on ability to pay (Monetary Authority of Singapore 2017). Such income-based limits and credit check requirements aimed to ensure that individuals do not borrow beyond their means. South Africa: Lenders must conduct an “affordability assessment” to understand a consumer’s financial situation before entering into a credit agreement (Republic of South Africa 2006). The assessment helps determine if the borrower can responsibly take on more debt. Credit providers must consider the following factors: debt repayment history, financial means, prospects and obligations, understanding of risks and costs, and rights and obligations under the agreement (see Appendix I). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 35 Section I Section II Section III Section IV Section V Appendix I Appendix II CHALLENGES A N D LIM ITAT IO N S requires greater effort by lenders and supervisors than As a principles-based approach, assessing the quality the more formalistic rule-based approach traditionally of product governance processes and practices prevalent in low-income settings. Solution 8: Data Protection Risk prevention • Data misuse WHY? (THE PA IN P O IN TS ) by growing concerns over the collection, use, and sale Digital lending models require intensive data gathering of consumers’ phone and internet data, for lending and sharing. Borrowers often do not know what or purposes. General data protection laws may be how data is being used and shared, nor can they easily enforced by a data or consumer protection authority. discover and control how lenders or their partners use this data. The customer data handling procedures of Data Protection Regulation specific to financial digital lenders and their partners have often proven services may be issued by such general agencies or inadequate, with numerous cases of failure to ensure by a financial regulator under the privacy provisions their due protection and security. Weak protection of financial laws. Regulation includes provisions on contributes to online fraud and client data misuse, disclosure, what data may be shared, consent by the including unauthorized charges, social engineering consumer to specific data uses, limitation of data uses to scams (e.g., phishing, spoofing), and social shaming. specifically agreed purposes, consumers’ access to their data, accuracy and security of the data, obligations of the data controller (e.g., the lender), and the consumer’s WHAT ? (THE SO LUT IO N S ) right to withhold or withdraw consent. Collaboration Data Protection and Privacy Legislation. Data laws among data, consumer protection, and financial have been widely adopted, including in emerging authorities is often critical to ensure effective regulation. markets and developing economies (EMDEs), driven Examples India: Under the Digital Personal Data Protection Act, 2023, digital “consent managers” positioned between entities involved in the data exchange manage the flows of information, registering consent, automating the exchange, and protecting confidentiality (Burman 2023, Ministry of Law and Justice 2023). A government board approves the consent managers and handles disputes. RBI’s Digital Lending Guidelines provide more specific standards, including restrictions on lenders’ uses of consumers’ mobile phone data (RBI 2022a). The Philippines: The National Privacy Commission’s guidelines on processing personal data for loans cover lenders and third parties. Mobile lending apps are barred from accessing personal information such as phone and social media contact lists (except for KYC purposes) to prevent unfair collection practices such as harassment and shaming. Lenders should provide just-in-time notices before obtaining consent and should provide information on how specified information will be processed (National Privacy Commission 2020) (see Appendix I). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 36 Section I Section II Section III Section IV Section V Appendix I Appendix II CHALLENGES A N D LIM ITAT IO N S data (Medine and Murthy 2020). Yet customers are Unlike conventional lending, digital credit is particularly frequently asked to agree to complex data uses and vulnerable to a wide range of cybersecurity risks, which flows they do not fully understand or cannot control. are expanding with the rapid adoption of smartphones As a result, reliance on consent alone offers limited and digital lending services. In many EMDEs, authorities protection. New approaches aim to balance customer often have limited capacity to identify and counteract data protection with the need for data to flow these risks. Consent is broadly seen as ineffective in throughout the financial system. protecting consumers or limiting firms’ use of their Solution 9: Combating Fraud Risk identification, prevention, mitigation • Fraud WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) The instant, automated, and remote nature of digital Regulatory provisions include rules on customer due credit, and the gaps in app security, augment fraud diligence, data protection, and security standards risks. Digital credit is more exposed than conventional applied to lenders, governance and risk management lending to the full range of consumer-harming cyber standards, and penalties related to financial fraud. frauds, such as: Market monitoring can help authorities detect significant risks or occurrences of digital credit-related • Social engineering scams, e.g., phishing or spoofing fraud that will inform regulations (see Section III, (fake SMS or websites that mimic legitimate lenders’ Solution 10). sites to harvest personal data), smishing or vishing (phishing via text or voice call), and QR code or Active collaboration among criminal investigators, mobile app frauds that promise easy access to prosecutors, and financial, data, and ICT authorities digital loans. is essential to curbing digital loan-related fraud and • Authorized push payment scams, i.e., when a other cybersecurity threats. This collaboration should fraudster tricks a consumer into sending money to a include staff capacity building and participation in criminally controlled account, for example, to pay a inter-institutional activities, an increase in information loan application or processing fee. exchange, and leveraging initiatives (e.g., ICT • Identity fraud, where a criminal gains control authorities’ SIM registration and authentication). of a consumer’s phone number or identity data through SIM swaps; biometric identity fraud where a fraudster breaches data storage to obtain copies C H AL L E N GE S AN D L I MI TAT I O N S As digital credit is subject to rapid innovation and of a consumer’s biometric data (e.g., fingerprints, change, fraudsters are often at the cutting edge and photos); or synthetic identity fraud where new pose a challenge to regulators who must keep pace identities are created by blending elements from to address new forms of fraud. Further, fraud spans different persons. Through identity fraud, fraudsters all digital activities, so financial authorities must move can then obtain new digital loans, renew them, or beyond their typical silos and engage with telco, data, increase their amounts. consumer, and judicial authorities. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 37 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples Malaysia’s National Scam Response Centre was established in 2022 as an operational center to coordinate rapid responses to online financial fraud. It involves the National Anti-Financial Crime Centre, the police, the central bank, the telco authority, FSPs, and telcos. Its hotline has received over 120,000 calls in two years, and its National Fraud Portal has helped FSPs and authorities rapidly identify, trace, and freeze stolen funds. In 2024, loan scams accounted for 15 percent of financial frauds reported to the police (Bank Negara Malaysia 2025; National Anti-Financial Crime Center n.d.). Since 2023, the Central Bank of Brazil has required all authorized FSPs (e.g., payment providers, fintech lenders) to collect and share data on occurrences and attempts of fraud in payment, deposit, and credit transactions among FSPs within 24 hours of their identification. FSPs must report an incident description, suspected fraudsters, the affected FSP, the recipient account, and the account holder (Banco Central do Brasil 2023). Solution 10: Digital Credit Market Monitoring Risk identification • All consumer risks and issues WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) Market monitoring shifts the focus of supervisory • Authorities can apply a range of quantitative (e.g., activities from individual lenders to the broader market analysis of regulatory reports, phone surveys) and borrower experience. The shift enables authorities or qualitative (e.g., mystery shopping, industry to gather early and in-depth insights into the borrowers’ engagement) tools and techniques that can be used experiences, risks, and outcomes associated with for several objectives (see Table 9). using digital credit; proactively identify consumer risks • First-time use of these tools helps create a baseline and issues that require greater attention; track market to better understand, identify, and measure developments, such as the use of new technologies consumer risks, vulnerabilities, experiences, and changes in borrower behaviors; and contribute to and results in digital credit. Periodic use allows disseminating supervisory expectations to the industry, monitoring, peer comparisons, and trend analyses especially to lenders that do not consistently receive that enable risk-based and forward-looking dedicated supervisory attention. supervision of digital credit. • In addition to authorities, consumer and research organizations (see Section V) as well as industry associations (see Section IV), can implement or pilot market monitoring tools, directly or in partnership with other ecosystem actors. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 38 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 9. Market monitoring tools can support different supervisory objectives Supervisory Analysis Analysis of Social Analysis Mystery Industry Thematic Phone Consumer objective of complaints media of shopping engagement reviews surveys advisory regulatory data monitoring consumer panels reports contracts Monitor indicators of consumer risk ✓ ✓ ✓ ✓ Monitor overindebtedness ✓ ✓ ✓ ✓ ✓ ✓ Monitor sales and marketing ✓ ✓ ✓ ✓ ✓ ✓ practices Monitor products in the market ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Monitor consumer complaints ✓ ✓ ✓ ✓ ✓ ✓ ✓ Monitor risks by gender ✓ ✓ ✓ ✓ ✓ ✓ ✓ Monitor consumer sentiment toward ✓ ✓ ✓ ✓ FSPs Monitor emerging consumer issues ✓ ✓ ✓ ✓ ✓ ✓ ✓ Source: Izaguirre et al. 2022a. CHALLENGES A N D LIM ITAT IO N S • Limited Depth: Market monitoring data is typically • Data Quality Issues and Resource Intensiveness: restricted to the formats, questions, and prompts Analyses of regulatory reports and complaints utilized at the tool design stage. Due to cost and demand strong efforts in adequately designing and resource constraints, they may have needed to developing reporting templates, using standardized undergo prioritization and simplification, but data definitions, and reviewing data submission and depth can be increased over time. validation, all in close collaboration with providers. • Unrepresentativeness: Tools like mystery shopping, Mystery shopping demands high-quality interviewer focus groups, consumer panels, and social media training to ensure that responses are adequately monitoring are not statistically representative. Still, recorded and not influenced by personal bias. they offer useful anecdotal insights into borrower • Compliance Costs: Regulatory and complaints experiences, risks, and outcomes. reporting requires a lot of data and may be too burdensome for authorities and lenders, especially For more information on phone surveys and social small ones with legacy systems. The more granular media monitoring, which may be commissioned to, or the data requested, the more costly the solution. carried out in partnership with research organizations, Regulatory technology (regtech) can help reduce see Section V. long-term costs, but also entails high upfront costs. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 39 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples Mystery Shopping: The European Banking Authority organized a shopping exercise with five authorities and thirty-seven FSPs to better understand the consumer experiences and risks of using a digital channel (or visiting a branch) to obtain a small personal loan (European Banking Authority 2023). The exercise revealed some lenders automatically increase the total credit amount to include fees without the consumer’s consent, do not provide adequate precontractual information, provide noticeably less information on digital channels than in branches, or rarely provide product-specific information via online chat. Supervisors were advised to investigate further, discuss with lenders, and issue guidelines. Analysis of Transaction Reporting Data: The Bank of Tanzania was concerned about the rapid growth in consumer credit, including digital. To address overindebtedness concerns, they worked with CGAP, Financial Sector Deepening (FSD) Africa, and FSD Tanzania, to pilot an ad hoc request of transactional and demographic data from digital lenders (CGAP 2022b). They gathered data on over 20 million loans disbursed over almost two years and identified a range of consumer risks, which informed regulatory improvements. Solution 11: Fraud Victim Compensation Risk mitigation • Fraud WHY? (THE PA IN P O IN TS ) fraud, consumer protection, consumer finance, Despite preventive measures, borrowers may still fall or electronic payments—the latter because victim to scams and frauds that cause significant digital credit-related frauds are perpetrated via monetary losses. The digital sphere has seen the electronic payment systems and the associated emergence of new types of fraud that are associated communications networks. with unregulated and illegal lenders, which may • Mechanisms to create compensation funds for significantly harm consumers and generate mistrust fraud victims in select cases can help mitigate the in the digital credit market. It is key for authorities to effects on consumers and encourage digital lenders take action to mitigate the effects of this fraud on and other firms in the digital ecosystem to invest customers and preserve trust in the financial system. strongly in fraud prevention measures. WHAT ? (THE SO LUT IO N S ) C H AL L E N GE S AN D L I MI TAT I O N S • Regulations clearly establishing the legal The difficulty of holding fraudsters accountable makes responsibilities of digital lenders, third parties, it even more important for authorities to collaborate in and customers, and the methods by which fraud setting up mechanisms to allocate responsibility and victims will be compensated. These provisions funding for victim compensation—a heavy lift for most typically appear in clauses that apply broadly to resource-constrained countries. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 40 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples US: The Dodd-Frank Act provides for a Civil Penalty Fund as part of CFPB. The Fund’s governing regulation (the Consumer Financial Civil Penalty Fund Rule), adopted in 2013, empowers the CFPB to compensate victims of federal consumer finance law violations, including illegal online lending practices (National Archives 2013). UK: The Payment Systems Regulator now mandates full reimbursement of up to USD 112,000 for victims of authorized push payment (APP) frauds, including loan fee scams. The sending firm shares liability equally (50/50) with the receiving firm. The fraud victim must be reimbursed within five business days (with limited exceptions), although grossly negligent consumers may be denied compensation. Consumers may seek redress with the Financial Ombudsman if unsatisfied with the settlement (Payment Systems Regulator 2023). This mechanism superseded the Contingent Reimbursement Model Code governed by the Lending Standards Board from 2019 to 2024. Solution 12: Effective Redress Mechanisms Risk mitigation • Inadequate redress WHY? (THE PA IN P O IN TS ) with clear instructions on how a consumer may Redress is often inadequate or insufficient where digital submit a complaint. credit is delivered by mixed partnerships or where no • Maintaining detailed and up-to-date written records regulated provider is involved. Many complaints may of all complaints and making them available to the be dismissed based on technicalities due to complex supervisor. or unclear procedures. As a result, borrowers may face unsolved problems, including monetary and privacy While redress regulations are often not specific to losses, and remain affected by abusive practices that digital credit, some require that FSPs offer redress go unaddressed. through the same channel used to deliver the product. In some cases, existing regulations have been amended to address complaints in the digital ecosystem. WHAT ? (THE SO LUT IO N S ) The law or regulation may require FSPs to establish Legislation may also establish alternative dispute complaints handling policies, procedures, and systems resolution mechanisms for financial sector authorities, that allow for consumers to appeal. At a minimum, consumer protection agencies, or data protection these standards for internal complaints-handling units authorities—providing an external channel for handling include: complaints and appeals. • Setting a maximum number of days for resolving complaints. • Providing system access through a range of channels (including the one used to extend the loan), e.g., phone, email, website, social media, along Responsible Digital Credit: Frontier Solutions for Authorities and Providers 41 Section I Section II Section III Section IV Section V Appendix I Appendix II CHALLENGES A N D LIM ITAT IO N S Redress mechanisms in digital credit may prove inadequate for several reasons: • Limited knowledge about how and where to complain and resolve complaints. • Lack of appropriate channels for correcting errors in credit reporting. • Lack of recourse against some unauthorized activities, such as data sharing or abusive collection practices. • Confusion over responsible parties in some digital credit models involving several providers. Examples India: Digital lenders must ensure that they and their partners have a redressal officer, adequately disclose this information to borrowers, and resolve complaints within 30 days (RBI 2021). In the second instance, customers of FSPs and credit information companies may go to RBI’s Integrated Ombudsman Scheme, which has simplified its processes and enables customers to contact a centralized platform using several mediums (e.g., physical mail, calls, emails) in English as well as ten local languages. Australia: The Australian Financial Complaints Authority (AFCA) was established in 2018 as an independent, nonprofit dispute resolution scheme for lending, deposits, and other products (AustLII 2018). Credit licensees, credit representatives, and debt management firms must join this scheme. Brazil’s online dispute resolution platform, managed by the National Consumer Secretariat, enables consumers and businesses to resolve disputes remotely (UNCTAD 2024). Fintech lenders participate in this platform, used by the Central Bank to inform its work (see Appendix I). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 42 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 13: Prevention of Predatory Lending and Abusive Debt Collection Risk prevention, mitigation • Overindebtedness WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) Digital credit can increase the risk of borrower • Regulations that identify cases where the lender is overindebtedness (see Figure 3), especially when responsible for granting credit in an irresponsible predatory or reckless business models and abusive manner. debt collection practices are in place. These issues • Regulations on responsible debt collection methods, increase consumer harm. While consumer credit outlawing social media shaming, harassment, threats regulations are generally designed to protect individual of violence, etc. retail customers, additional safeguards are often needed to ensure vulnerable digital credit consumers in low-income settings are adequately protected. C H AL L E N GE S AN D L I MI TAT I O N S Lenders’ incentives for misconduct are strong, while victims’ motivation to complain is weak, and authorities’ resources for investigation and enforcement are often inadequate. Further, external debt collectors are often outside the financial regulatory perimeter. Examples Predatory Lending: South Africa’s National Credit Act prohibits reckless lending, i.e., a loan where the lender fails to conduct an affordability assessment or enters a credit agreement even though evidence indicates that: i) the consumer did not generally understand or appreciate the risks, costs, or obligations under the proposed credit agreement; or ii) entering into that credit agreement would make the consumer overindebted. Courts can suspend or restructure reckless loans or refer them to a debt counselor for informal settlement (Republic of South Africa 2006). Predatory Lending and Debt Collection: Uganda’s Digital Lending Guidelines prohibit lenders from charging a default interest penalty that exceeds half the initial interest rate at the time of the loan offer (Uganda Microfinance Regulatory Authority 2024). They also prohibit improper debt collection tactics and conduct that harasses, oppresses, or abuses any person in connection with the collection of debt. FSPs must also identify their collection agents in advance (see Appendix I). Debt Collection: In Nigeria, the Central Bank—along with the ICT, anticorruption, and consumer agencies—formed the Joint Regulatory and Enforcement Task Force to rein in digital credit apps engaged in abusive debt collection practices and use of customer data, as well as charging unacceptably high rates to customers (FCCPC 2021). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 43 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 14: Personal Insolvency Risk resolution • Overindebtedness WHY? (THE PA IN P O IN TS ) C H AL L E N GE S AN D L I MI TAT I O N S Insolvency occurs when the individual (or solely owned Although bankruptcy regimes have been established MSE) is unable to pay their debts as they become for businesses in many jurisdictions, there is limited due because they do not have money available for experience with effective debt resolution regimes repayment or their debts exceed their assets—i.e., they in EMDEs. Past experiences with debt forgiveness are insolvent. measures that affected the repayment culture in EMDEs have demonstrated the importance of developing well-structured personal insolvency WHAT ? (THE SO LUT IO N S ) regimes. Additional concerns include the limited • Debt advisors and counselors who evaluate the capacity of judicial officials to manage these newer, borrower’s situation and recommend the best more complex cases involving personal finance. actions to take at a personal finance level and with creditors. • Bankruptcy or insolvency regimes with simplified procedures for very small businesses or individuals, which usually entail a legal filing and court procedure that determines how an insolvent person works out unpaid obligations with creditors. (World Bank forthcoming.) • Interim ad hoc procedures to provide borrowers with a temporary reprieve from creditor action. Examples US bankruptcy law provides for an orderly resolution of insolvency claims for individuals (US House of Representatives 1978). It offers comparatively strong formal protections to debtors and a fair and open resolution process for creditors. An individual debtor may seek liquidation or debt restructuring. Upon filing, legal debt collection is paused. The accompanying rules on exemption (at state and federal levels) allow the debtor to retain a share of their property. The CFPB also published guides for borrowers on how to work with credit counselors and the risks of contacting debt settlement companies. India’s 2016 Insolvency and Bankruptcy Code includes a “Fresh Start” process for low-income individuals (Ministry of Law and Justice 2016).a Upon filing, it provides for an interim moratorium that halts any pending legal action toward debt recovery and forbids creditors from initiating any such action against the debtor. The process provides a time-bound framework for debtors to negotiate debt restructuring. The parties also have a right to appeal. a This part of the Code is not yet in effect due to legal challenges. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 44 Section I Section II Section III Section IV Section V Appendix I Appendix II A Call to Action for Supervisors, technological solutions, and gather feedback on regulatory and supervisory measures and areas for Regulators, and Policy Makers improvement. Government authorities can strengthen consumer • Collaborate with research organizations and protection by collaboratively developing customer- consumer representatives to expand capacity for centric solutions that address gaps, conflicts, identifying, preventing, mitigating, and resolving and ambiguities in the regulatory and supervisory risks to consumers. framework. This requires taking action across the following areas: As digital credit continues to evolve, • Ensure that the appropriate regulatory authority has authorities need to embrace more proactive, clear jurisdiction over personal digital credit, with necessary supervisory and enforcement powers. preemptive, and coordinated efforts to build a Responsible Digital Credit Ecosystem. • Adopt a customer-centric view of responsible lending into regulation and supervisory policies A Call to Action for Funders and manuals. This includes recognizing the different stages of the digital borrower journey, the characteristics and vulnerabilities of digital Funders can play a critical role in advancing borrowers (especially those who repay late), and the responsible digital credit, by supporting the types of risks and outcomes they may face. development and adoption of government-led solutions. Key actions include: • Use technology and both demand and supply-side tools to monitor the digital credit market. Market • Raise awareness among grantees about the monitoring will help identify and assess consumer importance of addressing consumer risks in digital risks and outcomes, and inform subsequent credit through a more holistic and customer-centric regulatory and supervisory responses. Collaboration lens, one that considers the borrower journey, with research organizations, consumer associations, provider life cycle, and risk management cycle. and other authorities will strengthen monitoring • Advocate for the implementation of regulatory or (e.g., phone surveys, social media monitoring, and supervisory measures by making them a condition mystery shopping). for their credit lines. • Increase collaboration between financial and • Share promising responsible digital credit solutions nonfinancial authorities to share insights and data from this guide with regulators, supervisors, and on consumer risks and market developments, policymakers, and provide technical and financial identify regulatory and supervisory gaps, and resourcing needed to implement solutions suited to develop coordinated approaches to respond to the local context. fraud, unfair practices, data misuse, and other • Emphasize the importance of adopting financial related concerns. consumer protection principles and good practices • Develop regulatory measures that go beyond (OECD 2022; World Bank Group 2017), including general consumer protection laws to address the outcomes-focused and customer-centric regulatory most pressing consumer risks in digital credit, and supervisory frameworks to authorities (e.g., particularly overindebtedness through tailored Boeddu et al. 2021; Duflos et al. 2024; Izaguirre actions. 2020; Izaguirre et al. 2022a). • Engage with digital lenders and third-party service providers to understand the solutions they already have in place, build regulatory capacity on advanced Responsible Digital Credit: Frontier Solutions for Authorities and Providers 45 Section I Section II Section III Section IV Section V Appendix I Appendix II • Facilitate coordination among local authorities and support peer learning and knowledge sharing among authorities from different countries. • Support consumer associations by funding efforts to enhance their engagement with authorities, contribute to identifying consumer risks in digital credit markets, and advocate for improvements in both regulation and provider practices. Support research and consumer organizations to pilot solutions that enhance market monitoring and generate insights on emerging risks, especially in understudied areas such as behavioral vulnerabilities, agent-related risks, network downtime, and fraud. Methods may include phone surveys, social media monitoring, and mystery shopping. • Support innovation initiatives, such as hackathons, tech sprints, and regulatory sandboxes, to surface, test, and scale new solutions to make digital credit more responsible and strengthen dialogue among authorities, lenders, and tech providers. • Support capability development initiatives that help authorities better understand the benefits, risks, and outcomes of digital credit, and how they can act to address risks. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 46 Section I Section II Section III Section IV Section V Appendix I Appendix II SECTION IV Frontier Solutions for Responsible Personal Digital Credit: Providers P ROVIDERS PLAY A PIVOTAL ROLE IN • Digital lenders, e.g., fintechs, banks, and MFIs that expanding access to responsible digital credit are at the forefront of designing responsible credit by combining traditional credit data with products and must integrate affordability checks, alternative data sources—such as mobile phone ethical AI/ML models, and real-time risk monitoring metadata, digital payment patterns, and behavioral into their operations. signals- and integrating emerging technologies like AI, • Technology providers, e.g., AI/ML developers, ML, cloud infrastructure, and biometric authentication. cloud infrastructure firms, cybersecurity companies, These tools are used to improve credit assessments, bigtechs that enable real-time fraud detection, personalize customer experiences, reduce costs, biometric authentication, and AI/ML-powered risk and monitor emerging risks. Yet, the speed of digital assessment, but must balance innovation with innovation and the increasingly interconnected nature compliance and accessibility (see Box 4). of financial ecosystems amplify risks of fraud, data • Telcos and payment providers that facilitate loan privacy, and overindebtedness, driven by intensive origination, disbursement, and repayments and use of personal data, immediacy of services, and play a key role in expanding responsible financial growing reliance on technology. These risks may not inclusion, especially in mobile-first markets. only harm borrowers, but also providers’ reputation, sustainability, and profitability. As a result, providers • Credit bureaus and alternative data providers have a vested interest in adhering to responsible that enhance credit scoring for thin-file borrowers, lending practices that protect consumers. Providers leveraging psychometric, transactional, and are increasingly expected to strengthen their value behavioral data analytics. proposition by delivering real-time services and engaging in cross-sector collaboration. This includes Providers are developing and deploying a range of ensuring interoperability across the ecosystem; technology-enabled solutions to address consumer aligning with global standards for responsible lending, risks in digital credit throughout the customer journey transparency, and data security; and complying with (see Table 10). This research shows solutions catering national regulations. Providers must also support open to personal digital credit, with a focus on EMDEs and APIs and standardized protocols to enable seamless mobile-first markets, although some may also be integration of responsible digital credit solutions across applicable to broader DFS. platforms. Key actors include: Responsible Digital Credit: Frontier Solutions for Authorities and Providers 47 Section I Section II Section III Section IV Section V Appendix I Appendix II As the digital credit market continues to evolve, BOX 4. Roles of Bigtechs the path forward lies on strategic collaboration among digital credit providers, financial authorities, Bigtechs in EMDEs can play different roles: technology providers, research organizations, and civil • Setting the rules through app store policies society to balance access, security, and transparency, and enforcement (e.g., Google Play, Apple App and empowers providers to build sustainable and Store) to ensure transparency, restrict high-risk trustworthy solutions that contribute to a responsible lending apps, mandate user consent, and limit digital credit ecosystem. access to sensitive user data. • Powering digital lending infrastructure, including cloud services, AI/ML credit scoring tools, fraud detection, and analytics (e.g., Amazon Web Services, Microsoft Azure, Google Cloud, Meta, IBM). • Providing embedded credit through e-commerce ecosystem or platform-native BNPL solutions (e.g., Amazon Pay Later, Tencent’s WeChat Pay Credit). • Offering digital credit directly or via lending platforms and partners (e.g., Ant Group’s Huabei, Tencent’s WeBank). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 48 Section I Section II Section III Section IV Section V Appendix I Appendix II Responsible Digital Credit Solutions by Providers TABLE 10. Provider solutions by phase of risk management cycle, consumer risk, and customer journey step Provider Solution Risk Management Cycle Consumer Risk Customer Journey and Issue Step Identification Prevention Mitigation Resolution Borrower authentication 1.  methods Fraud prevention and 2.  detection systems, and Sign-up, application, AI-driven analytics Fraud disbursement, servicing Fraud reporting, 3.  transaction alerts, and accessible borrower support 4. Data encryption Data misuse Application, agreement, Transparency and control 5.  Data misuse, Lack servicing of borrower data-sharing of transparency preferences and consent  learer credit 6. C Lack of Agreement (T&C, disclosures and policy transparency terms disclosure) updates AI-enabled chatbots 7.  Inadequate Servicing, recourse, to provide multilingual, redress, agents and grievance 24x7 borrower support Redundant failover 8.  systems, disaster Network Disbursement, recovery, and offline downtime repayment capabilities  nbiased algorithms 9. U Credit assessment, and inclusive data in AI- Unfair treatment pricing, agreement based credit models (terms disclosure) 10. Borrower-centric Behavioral Product design, product design and vulnerabilities onboarding user experience Positive friction, 11.  dynamic pricing, automated debt Credit assessment, monitoring Overindebtedness pricing, servicing, 12.  Industry standards repayment for responsible digital lending A Call to Action for Digital Lenders and Service Providers A Call to Action for Funders Source: Authors. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 49 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 1: Borrower Authentication Methods Risk identification, prevention • Fraud WHY? (THE PA IN P O IN TS ) solutions must be tailored to users’ connectivity, device Weak borrower authentication processes expose access, and digital literacy level: digital credit users to fraud and financial losses • Multifactor Authentication (MFA): Multiple layers at multiple stages of their journey, largely during of verification (e.g., app-, SMS- or hardware-based) onboarding, loan applications, and repayment. tailored to different user segments and integrated These risks are particularly evident in EMDEs, where into digital lending platforms. MFA includes reliance on mobile devices is high, and financial and one-time passwords sent via text, email, or voice, as technological literacy may be limited. well as app notifications. It balances inclusion and security when designed with local contexts in mind. WHAT ? (THE SO LUT IO N S ) • Biometric Authentication: Facial, voice, or Robust borrower authentication is critical to secure the fingerprint verification for secure and seamless digital credit life cycle, particularly during onboarding, access, particularly useful where digital literacy is loan disbursement, and repayment. Authentication low. Biometric authentication eliminates passwords Examples Latin America: Digital lenders are exploring voice biometric authentication to reach low-literacy and underserved segments. Key biometric authentication providers include CPqD, TOC Biometrics, and ValidSoft (Biometric Update 2021). Global Security Technology Providers: Lead vendors enable secure onboarding and account access with robust MFA solutions such as RSA SecureID (token-based identity verification) and YubiKey by Yubico (phishing-resistant hardware-based MFA). Scalable Platforms for Integration: Microsoft Azure Active Directory and Google Authenticator offer MFA solutions that digital lenders can integrate at scale (ProjectPro 2024). Government-Led Biometrics Authentication: India’s Aadhaar program enables biometrics authentication and electronic KYC for digital lenders, especially with the recent launch of the Unified Lending Interface (Biometric Update 2024a; Reserve Bank Innovation Hub [RBIH] 2024). Adaptive Authentication: PayJoy, for example, monitors borrower behavior and adjusts security dynamically and in real-time based on transaction amounts and location (MIT Sloan 2023). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 50 Section I Section II Section III Section IV Section V Appendix I Appendix II and phishing risks. Voice verification works well in • False positives from biometric errors can lock contexts with low bandwidth. legitimate users out, reducing trust. • Adaptive Authentication: Dynamically adjusted • Privacy concerns around biometric data collection, security process based on risk context—e.g., device storage, and regulatory compliance, especially in type, geolocation, fingerprinting, and transaction regions without robust protections. value. Adaptive authentication enables higher • Borrowers experience trade-offs, as overly security with less user friction. complex authentication reduces usability and access for low-literacy or first-time users. CHALLENGES A N D LIM ITAT IO N S • Cost limits adoption of advanced tools by small • Unreliable internet access hampers MFA uptake in FSPs. low-connectivity areas. • AI/ML-enabled fraudsters require increasingly advanced defensive tools to thwart them.  raud Prevention and Detection Systems, Solution 2: F and AI-Driven Analytics Risk identification, prevention • Fraud WHY? (THE PA IN P O IN TS ) • Real-Time ML Fraud Detection: ML models Digital lenders handle large volumes of borrower data monitor fraud during loan application, and real-time transactions, from disbursements to disbursement tampering, and repayment anomalies repayments. Unauthorized access to this data can lead in real time, such as loan stacking, first-time to fraud (e.g., phishing attacks), causing financial losses defaults, or SIM swaps. and reputational damage. In EMDEs, digital credit users • AI-Powered Data Analytics: Analyze vast amounts typically experience fraud during loan application (e.g., of credit, transactional, and behavioral data (e.g., loan stacking, synthetic identities, identity spoofing) and utility bills, social media, repayment history) to flag loan disbursement (e.g., SIM swap, account takeover). risky behavior and assess trustworthiness. WHAT ? (THE SO LUT IO N S ) K E Y F E AT U R E S O F E F F E C T I V E F R AUD To prevent and detect fraud, digital lenders are DE T E C T I O N S O L U T I O N S deploying AI-driven tools tailored to mobile-first and • Scalability: Handle high transaction volumes high-volume environments: without hindering performance. • Behavioral Biometrics and Anomaly Detection: • Customization: Tailor models to detect fraud Tools analyze user patterns such as typing speed, patterns specific to digital credit use cases. mouse movement, swipes, and navigation habits to • Explainability: Use explainable AI to ensure detect account takeover attempts and suspicious transparency in loan approvals, customer support behavior. decisions, and regulatory compliance. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 51 Section I Section II Section III Section IV Section V Appendix I Appendix II • Real-Time Blocking: Flag or block suspicious • High costs of investment, implementation of transactions before loan disbursement. technology infrastructures, and talent may make this solution unaffordable for smaller digital lenders. • Black-box models reduce the explainability of why CHALLENGES A N D LIM ITAT IO N S certain repayment or disbursement transactions • Incomplete or unstructured data limits model or applications were flagged, which can erode accuracy. customer trust. • False positives may flag legitimate transactions or • Fraudsters continually adapt and leverage social loan applications, undermining trust. engineering and AI/ML tools to exploit vulnerabilities • Privacy concerns arise when analyzing large in fraud detection systems. amounts of user data. Compliance with data privacy regimes (e.g., GDPR) is critical. • Integration with legacy systems can be technically complex. Examples Behavioral Biometrics for Fraud Prevention: Risk operations platforms like those developed by the data science company Feedzai, offer •  behavioral biometrics, transaction monitoring, and account creation solutions. For example, a Brazilian bank leader in solar panel and vehicle loans was able to protect its customers from fraud attacks using Feedzai’s account hacking module, which leverages behavioral biometric, device, and network information (Feedzai 2024a). Real-Time Fraud Detection: TrustDecision is a software-as-a-service company that offers a multilayer fraud detection •  system, combining real-time decision-making and offline data retrieval, helping lenders detect loan application anomalies and repayment fraud. For example, a cash loan platform in Indonesia used TrustDecision to improve fraud detection accuracy (TrustDecision 2024). Feedzai’s AI models monitor loan disbursement and repayment transactions in real-time, •  identifying unusual patterns such as first-time payment defaults, irregular repayment behaviors, and synthetic identity fraud (Biometric Update 2025; Feedzai 2024b). AI-Powered Data Analytics for Case Investigation: Feedzai’s ML models enable lenders to investigate suspicious loan repayment patterns, analyze •  borrower transaction histories, and trigger dynamic fraud scoring, helping to prevent repeated borrower misuse (Bento et al. 2022). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 52 Section I Section II Section III Section IV Section V Appendix I Appendix II Fraud Reporting, Transaction Alerts, Solution 3:  and Accessible Borrower Support Risk mitigation, resolution • Fraud WHY? (THE PA IN P O IN TS ) repayments not initiated by them, or unauthorized Borrowers in EMDEs are often vulnerable to fraud, access to their accounts, enabling faster response. including loan stacking, unauthorized transactions, • Inclusive and Responsive Customer Support: and identity misuse (e.g., loans granted to a borrower’s Chatbots and live customer service agents, impersonator). These incidents can lead to damaged with multilingual or voice-based options, ensure credit, repayment confusion, and financial losses. Many accessibility and help underserved borrowers report borrowers face barriers to reporting fraud or accessing and resolve fraud quickly, reducing borrower harm alerts, due to low digital literacy, limited connectivity, and information fatigue. or a lack of multilingual support—especially for people with disabilities or people who are unable to speak an official language. C H AL L E N GE S AN D L I MI TAT I O N S • Infrastructure gaps delay fraud alerts, especially in low-bandwidth areas. WHAT ? (THE SO LUT IO N S ) • Legacy systems limit seamless integration with Digital lenders are implementing proactive and modern fraud detection technologies. inclusive fraud mitigation and resolution tools: • Language barriers hinder customer communication • Multilanguage Fraud Reporting: Robust reporting and fraud alerts. channels (e.g., app, SMS, calls) allow borrowers to immediately flag suspicious loan activity and • False positives worsen customer experience, confirm or dismiss alerts. creating mistrust or confusion. • Real-Time Transaction Alerts: Automated SMS or • High cost of customer support tools can deter in-app alerts notify customers of loan disbursement, digital lenders with limited resources. Examples Digital Lenders Using Fraud Alert Tools: Branch, a DFS company, uses in-app analytics and real-time alerts to identify, monitor, and •  mitigate the effects of suspicious loan applications and loan stacking in India, Kenya, Nigeria, and Tanzania (Branch 2024). Tala, a digital credit provider, sends SMS-based alerts for every loan disbursement, repayment •  date, or account change in India, Kenya, Mexico, and the Philippines (Tala 2024). Fraud Alert Infrastructure Providers: Twilio, a cloud-based communication technology company, enables real-time SMS and in-app •  fraud alerts embedded in digital lending platforms (Twilio 2024a; 2024b). Zendesk, a cloud-based software-as-a-service company, powers customer support tools, such •  as multilingual chatbots, to improve fraud reporting and resolution (Zendesk 2024). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 53 Section I Section II Section III Section IV Section V Appendix I Appendix II • Privacy concerns, data protection laws, and cross- • AI-powered fraud tactics evolve rapidly, border regulations complicate fraud prevention, necessitating that lenders continually update and requiring digital lenders to navigate complex refine their fraud prevention strategies. compliance requirements. Solution 4: Data Encryption All risk management phases • Data misuse WHY? (THE PA IN P O IN TS ) • Encryption at all stages—in transit, at rest, and in Data breaches can expose sensitive borrower use—protects borrower information throughout information—such as IDs, credit scores, and the loan cycle, and across systems it secures data transaction histories—leading to fraud, identity exchanges with credit bureaus, payment gateways, theft, and breach of privacy laws. Weak encryption and fraud detection systems. undermines borrowers’ trust and makes digital lenders vulnerable to reputational and financial risks, particularly in low-resource settings. C H AL L E N GE S AN D L I MI TAT I O N S • Compliance with Data Privacy Requirements: Laws and regulations may be fragmented and vary WHAT ? (THE SO LUT IO N S ) by country and region. Robust data encryption tools are critical to protect • Cost Barriers: Small lenders may lack the financial, sensitive borrower information with lightweight human, and infrastructure resources to implement processing technologies optimized for lower-end advanced encryption protocols. devices commonly used in underserved markets: • Legacy systems may lack compatibility with • Transport Layer Security (TLS) encrypts data modern encryption protocols. in transit, specifically during loan application, • Infrastructure gaps in EMDEs can limit real-time disbursements, repayment, or API interactions. TLS encryption and detection, especially for mobile-first 1.2+ is widely used in digital credit platforms to lenders. protect borrower data during online transactions. • Advanced Encryption Standard (AES) protects data at rest on servers and in cloud environments. AES-256 is a globally recognized standard used to secure borrower identity, credit history, and repayment data. • End-to-end encryption ensures that only the intended recipient (the borrower) can access sensitive communications, such as notifications about loan approvals, disbursements, or repayment schedules. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 54 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples Cloud-based platforms such as Google Cloud, AWS, and Microsoft Azure offer financial-grade encryption protocols (e.g., TLS 1.2+, AES-256) that secure API calls, backend infrastructure, and data in transit as well as at rest for digital credit platforms (AWS 2024; Google Cloud 2023; Microsoft 2024) (see Appendix II). Cybersecurity and technology providers, such as Thales CPL, Entrust, and AWS KMS, offer scalable encryption and centralized key management tools suitable for digital lenders in mobile- first and low-bandwidth environments (Entrust 2024; Thales Group 2024; Wawad 2023). Lender applications M-Kopa, in Kenya, uses AWS infrastructure, which includes built-in encryption and key •  management services (Freshworks 2024). Kubo Financiero in Mexico uses robust data protection and encryption measures (Traders •  Union 2024). Klarna, a global lending platform, uses advanced encryption methods, including AES-256, to •  secure data at rest and in transit (Stripe 2024). Solution 5: Transparency and Control of Borrower Data Sharing Preferences and Consent All risk management phases • Data misuse, lack of transparency WHY? (THE PA IN P O IN TS ) • Self-Service Dashboards: Allow consumers to easily Many digital borrowers, especially in underserved or view data collected by lenders via lending apps, low-literacy markets, struggle to understand how their update their data sharing preferences, and request personal data is collected, used, or shared. Without data deletion or corrections, where necessary. clear details about consent, they often agree to data • Product-Embedded Consent Mechanisms: Allow policies without fully understanding the implications. borrowers to grant, deny, or modify permissions As a result, they have limited control over their data in real-time through user-friendly in-app prompts, to manage sharing preferences, which augments data ensuring informed and revocable consent. misuse risks and erosion of trust. • Data Minimization Protocols: Lenders limit data collection to only what is essential for loan WHAT ? (THE SO LUT IO N S ) processing and credit scoring, avoiding excesses. To strengthen responsible data practices in personal • Tools and options for users to control their data digital credit, providers are introducing targeted tools preferences and opt in or out of certain types of that put borrowers in control: data usage. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 55 Section I Section II Section III Section IV Section V Appendix I Appendix II CHALLENGES A N D LIM ITAT IO N S • Data privacy fragmentation creates compliance • Implementation Costs: Building user-friendly complexity for regional or cross-border operations. dashboards and granular consent tools may deter • Third-party risks when compliant lenders share smaller providers due to cost implications. data with less-regulated partners. • Digital literacy gaps hinder borrowers’ • Evolving AI models rely on expanding data pools, understanding of data rights, even with simplified increasing pressure on responsible usage and tools. disclosure. Examples Tech Platforms Offering User Control Tools: Google provides dashboards for users to manage data preferences globally and has implemented regulatory measures for digital lending apps on Android devices in Africa (Google 2024a) (see Appendix II). Digital Lenders Applying Ethical Data Practices: Tala publishes a Data Ethics Policy and enables borrowers to manage how their personal data is used and shared (Tala 2024). Telecom-Driven Consent: Safaricom has implemented data privacy measures across its services, including M-Shwari digital loans, adhering to Kenya’s data protection laws and limiting third-party data sharing unless legally required (Safaricom 2024a; 2024d). Solution 6: Clearer Credit Disclosures and Policy Updates All risk management phases • Lack of transparency WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) Misleading or incomplete loan information—such as Providers are adopting proactive disclosures and policy undisclosed annual percentage rates (APRs), hidden strategies to improve transparency and borrower fees, or vague repayment terms—can result in poor protection across digital credit apps: financial decisions, unmanageable debt, and borrower • Standardized and Simplified Loan Disclosures: harm. These issues are especially problematic in Help consumers understand the fees, interest low-literacy markets where users rely on mobile apps rates, repayment terms, and the implications of late with limited regulation or inconsistent standards. payments at the point of application, using mobile- friendly formats. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 56 Section I Section II Section III Section IV Section V Appendix I Appendix II • Real-Time Loan Cost Estimators: In-app tools C H AL L E N GE S AN D L I MI TAT I O N S provide borrowers with upfront cost estimates, • Limited Platform Enforcement: Disclosure rules including fees, interest rates, and penalties, to often apply only to apps in official stores, creating support informed decision-making before accepting gaps for apps downloaded via sideloading or third- a credit offer. party marketplaces. • Platform-Level Transparency Enforcement: Major • Post-approval Circumvention: Some lenders may app stores (e.g., Google Play, Apple App Store) are change terms after app approval, making ongoing implementing policies to screen digital lenders, enforcement difficult. restrict access to sensitive user data, enforce • Low Borrower Awareness: Many users are unaware maximum APR limits, and require full loan costs. of platform safeguards or how to identify compliant apps, especially where financial literacy is low. Examples Global Bigtech Platforms Setting Standards: • Google requires digital lending apps on Google Play to disclose all loan terms, including APR and repayment period, and bans apps with APRs over 36 percent in certain countries (Google 2024b) (see Appendix II). • Apple has updated its App Store rules globally to mandate clear disclosure of all loan terms, including APR and payment due date. To curb predatory lending, it set a maximum APR of 36 percent and a minimum loan term of 60 days (Apple 2024) (see Appendix II). Lender-Side Disclosure Innovation: Fintech company Jumo, in partnership with CGAP, experimented with different types of communications linked to a digital credit offer and witnessed that simple changes to how costs are presented reduced default rates from 29 percent to 20 percent. These findings were then integrated into Jumo’s messages offering new digital loans. This experience demonstrated how clearer communication can improve repayment behavior and support responsible lending (Mazer and McKee 2017). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 57 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 7: AI-Enabled Chatbots to Provide Multilingual, 24x7 Borrower Support All risk management phases • Inadequate redress, agents WHY? (THE PA IN P O IN TS ) or harassment, help lenders meet standards for Borrowers face a lack of access to real-time agent or communication and conduct. staff support, and in underserved regions, they may not speak or understand the main language used in lending apps. C H AL L E N GE S AN D L I MI TAT I O N S • High Development Costs: Building and deploying AI-enabled chatbots requires financial and technical WHAT ? (THE SO LUT IO N S ) resources. Chatbots can address language barriers as their • Data Privacy Concerns: Borrowers are increasingly multilingual support helps borrowers in diverse regions concerned about how their data is used or shared. understand loan conditions and repayment schedules • Integration with Legacy Systems: Digital lending and reduce reliance on limited customer support platforms often rely on outdated systems that are teams. Since they are available 24/7, chatbots also incompatible with modern AI models. provide automated escalation processes to resolve complaints faster and improve cost efficiency by • Data Limitations: AI-enabled chatbots rely on large, handling generic high-volume queries allowing agents high-quality data sets to provide relevant responses. to focus on more complex cases. Specific solutions Data gaps or poor-quality training data can lead to include: inaccurate or irrelevant responses that frustrate customers. • AI-Enabled Chatbots, providing multilingual, 24/7 first-line customer support and assistance for • Explainability in AI: Complex AI systems may resolving queries and complaints, while escalating not adequately explain the reasoning for certain complex issues to support officers or agents. responses or solutions provided by chatbots, creating additional concerns. • AI and ML Tools, which monitor agent and officer interactions and detect patterns of non-compliance Examples Conversational AI Platforms for Digital Lending: BankBuddy supports digital lenders with multilingual, AI-powered bots to reduce loan processing •  delays and automate FAQs (BankBuddy 2024). Zendesk and Genesys power chatbot tools that script conversations with borrowers to ensure •  respectful, real-time responses across channels (Zendesk 2023). Mobile-First Chat-Enabled Platforms: Safaricom’s Zuri AI chatbot offers 24/7 customer support and is available on Telegram and Meta’s Messenger. Consumers can use Zuri to learn how to apply for an M-Shwari loan. In the future, borrowers will also be able to use Zuri for more advanced loan-related queries that are currently handled through the M-Pesa app and USSD (Safaricom 2024e). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 58 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 8: Redundant Failover Systems, Disaster Recovery, and Offline Capabilities All risk management phases • Network downtime WHY? (THE PA IN P O IN TS ) I N D U ST RY BE ST P R AC T I C E S AR E : Network interruptions can delay loan disbursement and • Decoupling critical services (e.g., loan repayment confirmation, especially in low-connectivity disbursement, repayment processing) from areas where borrowers rely on quick access for noncritical services (e.g., promotional notifications) emergencies. A single system failure may disrupt to ensure core functionalities are always operational. multiple services in the digital credit chain, undermining • Adapting a multicloud strategy, distributing trust and operational continuity. workloads across multiple cloud providers to mitigate downtime risks. WHAT ? (THE SO LUT IO N S ) Digital lenders and fintech partners are addressing C H AL L E N GE S AN D L I MI TAT I O N S reliability gaps with infrastructure designs that minimize • Regional regulations may delay the implementation downtime and ensure service continuity: of robust cloud or failover tools. • Redundant Failover Systems: Automatically reroute • Third-party dependencies increase exposure to functions to backup systems during outages to external failures or service withdrawal. maintain uptime. • Limited infrastructure remains a persistent barrier • Disaster Recovery and Resilience Planning: Ensure in rural areas. data protection and operational recovery in the • Vendor lock-in limits the ability to shift providers if event of failures or breaches. one fails or becomes anticompetitive. • Offline Capabilities: Enable essential loan services • Security risks around data loss, integrity, and (e.g., application, repayment processing) through access must be mitigated through encryption and USSD SMS, or cached app functions when role-based permissions. connectivity is unavailable. • Transparent Outage Communication: Promptly notify customers about disruptions and expected resolution times via SMS or app notifications. Examples Digital Lenders with Failover Strategies: Branch and other digital lenders operating in EMDEs have implemented scalable infrastructure and redundant system designs to maintain loan service availability, especially during network disruptions. Cloud Infrastructure Providers, such as AWS, Google Cloud, Microsoft Azure, and IBM Cloud, power digital lenders like Tala and Branch, enabling scalable operations and high uptime for loan applications, disbursements, and repayments, as well as automating disaster recovery in low-connectivity environments (see Appendix II). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 59 Section I Section II Section III Section IV Section V Appendix I Appendix II  nbiased Algorithms and Inclusive Solution 9: U Data in AI-Based Credit Models All risk management phases • Unfair treatment WHY? (THE PA IN P O IN TS ) • Implementation of fairness testing and algorithm Consumers may face unfair lending practices. AI-based adjustment for credit scoring and loan approval credit scoring can unintentionally amplify biases if processes, and open-source fairness toolkits to models are trained on nonrepresentative data or rely test models for discriminatory patterns and adjust on variables that correlate with income, gender, or scoring models and weighting systems accordingly. geography, which can lead to discriminatory lending • Ethical automation and application of debt outcomes. For instance, algorithmic bias may prioritize collection standards, by using AI tools to monitor certain factors and assign more weight to elements agent behavior, automate reminders, and prevent correlated with protected classes or socioeconomic harassment—ensuring compliance with ethical and status. In EMDEs, this can result in unfair loan regulatory standards. rejections, pricing disparities, and unethical debt collection—eroding trust and excluding vulnerable consumers from financial systems. C H AL L E N GE S AN D L I MI TAT I O N S • The high cost and complexity of accessing and maintaining quality and diverse data sets to train AI WHAT ? (THE SO LUT IO N S ) models, and implementing fairness tools. Providers are advancing tools and practices that reduce • Evolving regulatory frameworks make it harder discrimination and promote responsible automation in to ensure AI systems and algorithms comply with digital lending: regulations and standards, including cybersecurity, • Use of ML and inclusive training data to assess to protect AI systems and data from cyberattacks creditworthiness and continue monitoring and and threats. auditing for bias of AI-based algorithms and • Explainability difficulties in “black-box” AI models AI-driven credit scoring. can reduce transparency and accountability. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 60 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples AI Fairness Toolkits: A growing suite of open-source toolkits enables FSPs to assess and mitigate algorithm bias. • Microsoft Fairlearn: Developed as a Python library, Fairlearn is designed to support fairness-aware ML practices. It provides tools for measuring disparities in model outcomes across demographic groups, applying fairness constraints during model training, and reducing algorithmic bias through post-processing reweighting strategies. In credit scoring, Fairlearn enables developers and data scientists to evaluate and address biases that may disproportionately impact protected groups such as women, rural populations, or ethnic minorities. Its interpretability functions also support transparency and model auditability, consistent with emerging regulatory expectations (Microsoft 2020) (see Appendix II). IBM AI Fairness 360 (AIF360): AIF360, available in both Python and R, is a comprehensive, •  open-source toolkit offering fairness metrics and a wide range of bias detection and mitigation algorithms. Relevant use cases in financial services include loan approval, credit risk assessment, and interest rate determination, where AIF360 enables institutions to test for disparate impact and adjust models accordingly. Google’s What-if-tool (WIT): WIT is an interactive visualization interface that integrates with •  TensorBoard and TensorFlow models. It enables users to explore counterfactual scenarios, test the sensitivity of model predictions to changes in input variables, and simulate model performance across subgroups. For example, a lender can visualize how changing an applicant’s income or location affects loan approval probability, helping to identify unintended biases. Responsible Lender Practices: Tala uses AI and behavioral and transactional data—including smartphone usage patterns, SMS metadata, and repayment behavior—to customize credit limits and repayment schedules based on borrowers’ demonstrated capacity to repay. This approach extends credit to “thin file” customers without traditional credit histories. Tala has published a Data Ethics Policy outlining its commitment to responsible AI use. As part of its practices, Tala explicitly excludes sensitive personal attributes—such as gender, race, ethnicity, religion, gender, and political affiliation—from its credit scoring models (Tala n.d.). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 61 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 10: Borrower-Centric Product Design and User Experience All risk management phases • Behavioral vulnerabilities WHY? (THE PA IN P O IN TS ) • Product Design and Real-Time Monitoring: Poorly designed digital lending products can lead FSPs implement backend systems to address to negative friction, resulting in confusion and performance issues, service glitches, payment a frustrating user experience. Consumers may failures, and user friction in real-time, enabling misunderstand T&C, encounter errors, or face proactive problem resolution. failed payments. Many apps are not designed to • In-App Reporting Enhancements: They provide accommodate consumers with disabilities or low digital clear, actionable borrower information, including literacy. Technical glitches and system downtime can repayments status, due dates, outstanding balances, further disrupt access and repayment. Further, lack and credit histories, to promote financial awareness. of transparency and explainability—with complex • Algorithm Explainability: FSPs integrate explainable algorithms and data-driven decisioning processes— (XAI) practices that demystify lending decisions, can also make it difficult for consumers to understand enabling borrowers to understand why they qualify the rationale behind lending decisions. (or not) for loans in plain language. WHAT ? (THE SO LUT IO N S ) • Customer-Centric Interfaces: Credit providers C H AL L E N GE S AN D L I MI TAT I O N S • Costs: Lenders may face high implementation costs that prioritize customer-centricity offer seamless, for upgrading security and complying with varying intuitive, and mobile-first user experiences that data protection laws and regulations. work across feature phones and smartphones, accommodating varying digital literacy levels. • Balanced Approach: Lenders need to balance seamless user experience and convenience with • Omnichannel Access: FSPs ensure borrowers can robust security and data protection. interact with digital credit products through multiple channels—apps, SMS, USSD, or call centers—to improve inclusivity and continuity. Examples M-Kopa piloted an interactive SMS system that allowed borrowers to access their credit histories for free and notify the credit bureau of any inaccuracies. Borrowers who opted into the SMS system took up more credit but were more likely to repay on time (Mazer and McKee 2017). US Cash App redesigned its interface to include voice guidance, larger fonts, and simplified navigation, enhancing usability for elderly and visually impaired users (Cash App 2024). Klarna has implemented explainable AI algorithms to help users understand loan approval or rejection decisions, building trust among borrowers (Klarna 2023). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 62 Section I Section II Section III Section IV Section V Appendix I Appendix II  ositive Friction, Dynamic Pricing, Solution 11: P and Automated Debt Monitoring All risk management phases • Overindebtedness WHY? (THE PA IN P O IN TS ) • Automated debt monitoring tools that integrate Greater ease of borrowing through digital credit with credit bureaus or other sources to provide products can increase the risk of overindebtedness borrowers with real-time updates on their debt for low-income customers, including people prone status and repayment progress. to gambling. Some digital lenders also extend or roll over loans without reassessing repayment capacity, compounding debt stress. High interest rates and C H AL L E N GE S AN D L I MI TAT I O N S • Digital lenders often optimize for speed over lack of transparency further increase vulnerability, affordability and risk assessment. particularly in low-literacy markets. • Borrowers may lack the financial literacy to understand how high-interest rates or penalties can WHAT ? (THE SO LUT IO N S ) exacerbate debt and negatively impact their credit • Positive Friction by Design: Intentionally scores. introducing steps at strategic points in a consumer • Fragmented and inconsistent or absent regulations journey to slow users down and shift them out of on digital lending in certain regions allow unethical automatic thinking, allowing them to make more practices to persist. informed decisions. • Dynamic pricing and flexible repayment based on borrower income and credit history. This aligns repayment schedules with the ability to repay and reduces default risk. Examples Positive Friction: In Kenya, Jumo and CGAP tested positive friction by making loan terms harder to skip. This increased borrower engagement and improved repayment rates, showing that simple design changes can encourage more responsible borrowing (Mazer and McKee 2017). Dynamic Pricing and Flexible Repayment: Branch offers flexible repayment schedules with terms ranging from 4 to 12 weeks in countries like Kenya and Nigeria and provides incentives for early repayment through increased loan limits for future borrowing. Tala dynamically adjusts both credit limits and interest rates based on borrower capacity, using risk-based pricing that analyzes over 250 alternative data points, including repayment behavior, phone usage patterns, and transaction history to create personalized loan offers (Branch n.d., Tala 2021). Automated Debt Monitoring: Plaid’s Liabilities API lets lenders track repayment history, debt balances, and upcoming due dates—enabling early intervention and borrowers’ visibility of loan information (Plaid 2024). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 63 Section I Section II Section III Section IV Section V Appendix I Appendix II  ndustry Standards Solution 12: I for Responsible Digital Lending All risk management phases • Overindebtedness WHY? (THE PA IN P O IN TS ) report cases of noncompliance, sanctions, or Greater ease of borrowing through digital credit can removal from the association. increase the risk of overindebtedness for low-income • Self-Regulatory Role: Authorities may assign customers, especially customers who are prone to associations’ responsibilities not only to develop and gambling. Customers who borrow with limited need or enforce the code, but also to register digital lenders intention have an increased risk of nonrepayment and and become a self-regulatory organization (see overindebtedness. Some lenders extend or roll over Section III, Solution 1). loans without evaluating the borrower’s ability to repay, • Market Monitoring: Associations also produce increasing the overall debt burden. industry reports and statistics that help identify opportunities, risks, and issues. WHAT ? (THE SO LUT IO N S ) • Codes of Conduct: Industry associations develop, issue, and require member adherence to codes of C H AL L E N GE S AN D L I MI TAT I O N S • Industry associations may require time to develop conduct. The code includes provisions on aspects the capacity and resources necessary to monitor such as thorough affordability assessments, fair and enforce compliance with codes of conduct. pricing, transparency, and complaints handling. Some may also address financial coaching, training, • When membership is voluntary, an association may and counseling. Associations may work with not be able to include all relevant actors, or multiple authorities to develop codes. competing small associations may emerge with differing standards. • Enforcement Mechanisms: Industry associations list members who endorse the code and have a • Collaboration with authorities may be challenging mechanism to monitor and enforce it. They also due to limited resources, unfamiliarity, or concerns about regulatory capture. Examples India: The Fintech Association of Consumer Empowerment (FACE) issued a code of conduct for digital lenders, covering areas such as transparency, pricing, collections, data usage and sharing, and grievance redressal. Members are required to endorse the code and may be sanctioned for noncompliance. FACE also monitors the market and flags illegal loan apps to Google for removal from their Play Store (FACE n.d.). Singapore: The Singapore Fintech Association (SFA) issued a code of conduct for BNPL providers in October 2022, covering disclosure, marketing, redress, and financial hardship assistance. Members can be excluded from the list of registered providers due to noncompliance (SFA 2022). Kenya: The Digital Financial Services Association (DFSAK) issued a code of conduct for responsible lending in 2021, covering credit assessment, disclosure, marketing, debt collection, and third-party arrangements. It is enforced by a Disciplinary Committee, which may suspend or cancel membership due to noncompliance (DFSAK 2019). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 64 Section I Section II Section III Section IV Section V Appendix I Appendix II A Call to Action for Digital • Prioritize customer-centric digital credit solutions Lenders and Service Providers by embedding transparency into product design, clearly disclosing loan terms and data use practices, Digital lenders and digital credit service providers can and equipping borrowers with tailored tools, leverage technology to prioritize customer-centric resources, and financial education to support solutions and innovate in responsible digital lending. informed decision-making and responsible They can: repayment. • Foster ecosystem-wide connectivity by aligning • Strengthen security and data protection by regulatory frameworks, adopting open APIs and adopting robust, seamless authentication methods, shared data standards, and enabling seamless encryption protocols, and secure handling of cross-sector partnerships among lenders, fintechs, borrowers’ transactional and alternative data across bigtechs, and mobile platforms. This ensures the customer journey. scalable, secure, and inclusive delivery of digital • Harness emerging technologies to drive scalability, credit while reinforcing transparency, regulatory efficiency, and resilience by implementing compliance, and user trust. Support innovation responsible AI/ML–powered solutions such as initiatives, such as hackathons, tech sprints, inclusive credit scoring models, advanced and and regulatory sandboxes, to surface, test, and real-time fraud detection systems, and smart scale new solutions to make digital credit more chatbots that enhance borrower engagement. responsible and strengthen dialogue among authorities, lenders, and tech providers. • Raise awareness among investees and grantees about the importance of addressing consumer risks in digital credit through a more holistic and Providers have a critical role to play in customer-centric lens, one that considers the building consumer trust by offering credit borrower journey, provider life cycle, and risk products and services that deliver value management cycle. without causing harm. By prioritizing • Promote responsible lending practices through consumers, and fostering long-term loyal enforceable codes of conduct (particularly in early stages of market maturity), positive friction, dynamic engagement, providers not only enhance risk-adjusted pricing, and real-time debt monitoring. customer value but also strengthen business value, growth, and sustainability. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 65 Section I Section II Section III Section IV Section V Appendix I Appendix II TABLE 11. Implementing responsible digital credit solutions in EMDEs requires careful trade-offs Provider Solution Trade-offs and Balances Borrower authentication methods Balance strong security with simplicity and accessibility, especially for first-time users, elderly borrowers, and people with disabilities or low digital literacy. Fraud prevention and detection systems, and Ensure informed consent, data minimization, and alignment with privacy laws AI-driven analytics for biometrics. Fraud reporting, transaction alerts, and Balance security with user experience for behavioral biometrics. accessible borrower support Manage the tension between real-time fraud detection, explainability, and affordability, especially in low-resource settings. Balance fraud-related actions with customer experience. Data encryption Implement robust protections while maintaining system affordability and compatibility with any existing legacy infrastructure. Transparency and control of borrower data- Balance transparency and complexity, ensuring lenders do not overwhelm sharing preferences and consent customers with too much detail or prioritize quick user acquisition and Clearer credit disclosures and policy updates lending growth. AI-enabled chatbots to provide multilingual, Ensure support is accessible, clear, and concise to avoid overwhelming 24/7 borrower support customers, keeping them informed and protected. Redundant failover systems, disaster recovery, Ensure service continuity without dramatically increasing operational costs, and offline capabilities particularly in rural or low-bandwidth regions. Unbiased algorithms and inclusive data in AI- Tailor fairness toolkits with contextual borrower insights, for example, digital based credit models behavior and local norms, to avoid reinforcing bias while preserving model performance. Borrower-centric product design and user Build products that empower informed borrower choices through intuitive experience and mobile-friendly design. Source: Authors. A Call to Action for Funders • Provide financial and nonfinancial support to lenders and those providing responsible lending solutions, Funders can support the development and adoption of demonstrating to the market that there are solutions led by providers. They can: business opportunities to invest in these solutions. • Raise awareness among grantees and investees Also, provide incentives (e.g., reduced interest or about the importance of addressing consumer performance-based rewards) to investees that risks in digital credit through a more holistic and adopt responsible lending practices. customer-centric lens, one that considers the • Before lending to or investing in digital lenders, borrower journey, provider life cycle, and risk including BNPL providers, conduct adequate due management cycle. diligence on their responsible lending practices and • Share promising responsible digital credit solutions processes to ensure they have adopted or plan to from this technical guide with providers and offer adopt solutions to prevent, identify, mitigate, and technical and financial resources to implement resolve consumer risks. Pay special attention to solutions suited to the local context. their culture and conduct governance, conduct risk management policies and systems, and financial Responsible Digital Credit: Frontier Solutions for Authorities and Providers 66 Section I Section II Section III Section IV Section V Appendix I Appendix II product governance procedures that underlie adequate responsible lending practices (Izaguirre 2020). • Monitor how digital lenders are implementing client protection principles (e.g., Cerise+SPTF 2024; OECD 2022; World Bank Group 2017) and solutions to address consumer risks, including solutions already implemented elsewhere. • Support hackathons, tech sprints, etc., to surface and test new solutions with tech providers to make digital credit more responsible. • Promote collaborations among providers, research organizations, or consumer associations to pilot solutions that address overindebtedness, and contributing factors such as behavioral vulnerabilities, and facilitate authority-provider dialogue and peer learning. • Support the setup, strengthening or reforms of actors that can help make digital credit more responsible, especially when their services account for the needs and complexities of an evolving digital credit market (e.g., credit reporting institutions, debt counsellors). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 67 Section I Section II Section III Section IV Section V Appendix I Appendix II SECTION V Frontier Solutions for Responsible Personal Digital Credit: Research Organizations and Consumer Representatives Role of Research Organizations These include market monitoring tools, credit scoring algorithms, and consumer communication strategies in Promoting Responsible that show promising results in tackling unfair treatment, Digital Credit transparency gaps, and behavioral vulnerabilities in digital credit. Research organizations and networks play a critical role in identifying, preventing, and mitigating Despite this progress, challenges still remain. Limited consumer risks in digital credit. By analyzing customer access to data, funding constraints, and complex experiences, market practices, and emerging regulatory environments can hinder long-term research threats, they generate evidence that helps shape and stakeholder collaboration. Addressing these policies and tools to address issues such as fraud, barriers is important to ensure that research continues overindebtedness, and data misuse. to inform responsible innovation and consumer protection in digital credit. Through partnerships with providers and authorities, research organizations such as CEGA, CGAP, and IPA have developed and tested practical solutions. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 68 Section I Section II Section III Section IV Section V Appendix I Appendix II Responsible Digital Credit Solutions by Research Organizations TABLE 12. Research organization solutions by risk management cycle and consumer risk Research Organization Solution Risk Management Cycle Consumer Risk and Issue Identification Prevention Market monitoring—partnering with authorities 1.  to conduct phone surveys All consumer risks and issues 2. Market monitoring—developing methodologies to analyze social media using artificial intelligence Market monitoring—using mobile app metadata 3.  Fraud to detect fraudulent apps Machine learning models to predict 4.  Overindebtedness overindebtedness Piloting gender-differentiated algorithms to 5.  Unfair treatment improve women’s access to credit  ommunication strategies and formats to 6. C Lack of transparency, Behavioral increase transparency vulnerabilities Market Monitoring—Partnering Solution 1:  with Authorities to Conduct Phone Surveys Risk identification • All consumer risks and issues WHY? (THE PA IN P O IN TS ) this data to complement providers’ regulatory reports Research organizations and networks often set their and gain a deeper understanding of emerging issues. own research agendas, which can support market monitoring, typically under the purview of market conduct supervisors (see Section III, Solution 10, and W H AT ? ( T HE S O L U T I O N S ) Phone surveys are a cost-effective and efficient way CGAP’s Market Monitoring Toolkit) (Izaguirre et al. to collect direct borrower data, offering insights into 2022a). Researchers can play an important role because market trends, financial behaviors, and emerging risks. they serve as neutral third parties that collect demand- These surveys help monitor consumer issues such as side data, providing insights into customers’ financial unfair practices and overindebtedness, while enabling experiences and the reasons noncustomers avoid demographic segmentation. financial services. Market conduct supervisors can use Responsible Digital Credit: Frontier Solutions for Authorities and Providers 69 Section I Section II Section III Section IV Section V Appendix I Appendix II CHALLENGES A N D LIM ITAT IO N S ownership among women and limited coverage in Phone surveys face challenges related to sample rural areas. Additionally, the quality of responses may representativeness and data quality. Mobile phone be compromised, as interviewers are unable to rely on access remains uneven, with around 30 percent nonverbal cues to build rapport or sustain engagement. of the global population lacking access, and lower Examples IPA implemented random-digit-dial remote phone surveys in Kenya, Nigeria, and Uganda, in collaboration with national authorities, to surface consumer protection priorities among users of digital finance, including digital credit (IPA 2023a). CGAP developed digital credit phone survey questionnaires that were piloted in Kenya and Tanzania, in collaboration with national authorities, to uncover consumer protection risks among digital credit users differentiated by gender and other segments (CGAP 2022a; Kaffenberger 2018).  arket Monitoring—Developing Methodologies Solution 2: M to Analyze Social Media Using Artificial Intelligence Risk identification • All consumer risks and issues WHY? (THE PA IN P O IN TS ) digital credit apps that lack formal redress mechanisms. With the rise of DFS, consumers are increasingly Additionally, financial supervisors can use NLP to turning to social media to share their experiences and monitor unregulated digital finance providers by complaints. Analyzing this content can provide timely analyzing social media content. and valuable insights into challenges with digital finance, including those related to new products and emerging providers. Social media monitoring can C H AL L E N GE S AN D L I MI TAT I O N S NLP analysis of social media faces challenges in serve as a useful complement to traditional survey multilingual countries where local slang, dialects methods, offering more frequent and direct access to and informal language complicate accurate text consumer perspectives. interpretations. Additionally, data cleaning is critical, requiring sophisticated methods to filter through WHAT ? (THE SO LUT IO N S ) noise and ensure quality, as unstructured text is rich in Research organizations can leverage natural language irrelevant or misleading information. processing (NLP), a type of AI, to analyze social media content at scale. NLP helps detect complaint trends in social media by processing large volumes of text. It also captures consumer concerns, especially for Responsible Digital Credit: Frontier Solutions for Authorities and Providers 70 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples IPA, in collaboration with AI provider Citibeats, piloted a social media listening and analysis tool for consumer protection monitoring in DFS, including digital credit, in Kenya, Nigeria, and Uganda (IPA 2023c). They collected historical data on consumer protection-relevant content from Twitter, Facebook pages, and Google Play Store reviews, and analyzed them using NLP approaches. In a pilot project, CGAP and Decodis utilized NLP to analyze over 150,000 social media posts and Google Play reviews related to digital lending apps in India. The study identified over 25,000 complaints, with significant concerns about aggressive debt collection and fraudulent apps, showing how NLP-based social media monitoring can serve as an effective early warning system for consumer risks in digital lending (Duflos et al. 2021a). In 2022, for the second round of the pilot, the RBIH joined the exercise and utilized the results (Duflos et al. 2023). Solution 3: Market Monitoring—Using Mobile App Metadata to Detect Fraudulent Apps Risk identification • Fraud WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) Predatory and fraudulent practices in digital finance, To address these risks, research organizations can particularly through personal loan apps, have risen test market monitoring tools that analyze mobile app globally over the past decade. The widespread metadata and consumer complaints data to identify adoption of mobile devices, coupled with low financial and flag suspicious digital finance apps. These tools and digital literacy among many consumers, has can offer an early warning mechanism for authorities made vulnerable individuals and households easy and consumer advocates. targets. Evidence of exploitative behavior by some app providers has increased concerns about consumer harm and privacy violations. These practices not only C H AL L E N GE S AN D L I MI TAT I O N S A key limitation of the IPA market monitoring cause harm, but also erode trust in digital finance, pilot below was its exclusion of non-English apps. potentially slowing financial inclusion progress. Addressing this gap would require developing methods to incorporate text-based analysis or NLP tools adapted to multiple languages. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 71 Section I Section II Section III Section IV Section V Appendix I Appendix II Examples IPA piloted a market monitoring tool using mobile app metadata, user reviews, and data downloaded from digital finance apps on the Google Play Store, combined with ML, to flag suspect finance apps. Covering 134,744 unique app IDs across 63 countries from January 2020 to April 2021, the tool analyzed both visible and hidden app store data. Focusing on personal loan apps, IPA tested two labeling approaches: 1) Manual classification using heuristic techniques and ML models, and 2) Market-specific guidance based on lending regulations in India, Nigeria, and the Philippines. Both approaches were benchmarked against actual app removals to assess effectiveness. Results indicate that the ML models achieved high (80–90 percent) accuracy for binary classification and fair (70–80 percent) accuracy for multi-class predictions, even using only static input features (Fu and Mishra 2022). Solution 4: Machine Learning Models to Predict Customer Overindebtedness Risk identification, prevention • Overindebtedness WHY? (THE PA IN P O IN TS ) ML models designed to predict borrower distress. Debt distress or overindebtedness pose significant These models use readily available data, such as past challenges, especially for low-income borrowers who repayment behavior, demographic characteristics, and often have irregular income streams and limited capacity economic activity, to determine borrower distress early.  to absorb financial shocks. Without early detection and intervention, low-income households may resort to harmful coping strategies, such as reducing essential C H AL L E N GE S AN D L I MI TAT I O N S Despite their promise, ML models may struggle to expenditures on education and healthcare, to avoid loan differentiate between temporary financial strain default and maintain access to credit. and long-term distress, leading to false positives or negatives. In addition, black-box models make it WHAT ? (THE SO LUT IO N S ) difficult for regulators, providers, and borrowers to To address this, research organizations have understand why certain decisions are made. collaborated with FSPs to develop and test AI and Examples In India, Dvara Research, in collaboration with the Robert Bosch Center for Data Science and AI at IIT Madras, has developed an ML model designed to predict distress among borrowers caused by debt. The goal of the project was to identify households at risk of overindebtedness before defaults occurred. The project has tested the model with over 1.5 million customers of a nonbank lender, and validated the results using field surveys (Dvara Research 2023). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 72 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 5: Piloting Gender-Differentiated Machine Learning Models to Improve Credit Access for Low-Income Women Risk prevention • Unfair treatment WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) Credit algorithms can unintentionally disadvantage Gender-differentiated credit scoring algorithms can low-income women due to biases in both data address these gaps, by incorporating variables that collection and model design. Traditional credit scoring better reflect women’s socioeconomic characteristics systems often rely on data points such as formal and financial behaviors, particularly in low-income employment records or property ownership, which settings. These tailored models can more accurately many low-income women may not have. This absence assess creditworthiness and ensure that women who of relevant data points can lead to lower credit scores, may be unfairly rejected by traditional scoring methods limiting their access to financial services. Additionally, if (such as pooled models) gain fair access to credit. algorithms are trained on data sets that predominantly reflect typical male financial behaviors, they may not accurately assess the creditworthiness of women, C H AL L E N GE S AN D L I MI TAT I O N S Due to the sensitive nature of the data collected, the leading to unfair loan rejections or higher interest rates. pilot in the example below required careful setup of secure data infrastructure, including anonymization and encryption, to comply with local privacy regulations. Examples Researchers from CEGA partnered with a bank in the Dominican Republic to pilot a credit scoring model by analyzing data from 16,091 credit card applicants. They also collaborated with the telco Claro to incorporate call detail records and additional sociodemographic data to build a gender- differentiated credit scoring model. The study simulated three credit scoring models (traditional, male-only, and female-only) and found that over one-third of women rejected by the traditional model would be approved by a gender-differentiated model. Preliminary results suggest that about 80 percent of women receive higher credit scores with a gender-differentiated model, which incorporates additional data, including mobile phone usage (CEGA 2023). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 73 Section I Section II Section III Section IV Section V Appendix I Appendix II Solution 6: Communication Tools and Formats to Increase Transparency Risk prevention • Lack of transparency, behavioral vulnerabilities WHY? (THE PA IN P O IN TS ) W H AT ? ( T HE S O L U T I O N S ) A lack of transparency surrounding T&C of digital In partnership with FSPs, CGAP has conducted credit products, due to inaccessible, complex, or pilots to test communication tools and formats to unclear information, is a common risk for customers. improve transparency in digital credit products. These risks are often exacerbated by provider These include interactive SMS messages to enhance practices that exploit behavioral vulnerabilities, such engagement with consumers, and the redesign of as framing messages in ways that are designed digital loan application processes to promote clearer to influence consumer choices. While numerous understanding of loan T&C. solutions tackle these risks from various angles (often regulatory), researchers can also play a role through innovation and testing. C H AL L E N GE S AN D L I MI TAT I O N S While certain behavioral messaging strategies can have a positive impact on repayment for subsegments of digital loan borrowers, they need to be rigorously tested and targeted. Examples CGAP partnered with Busara and Jumo in 2015 to design a series of experiments in Kenya to measure consumer responses to different communications about a digital credit offer (Mazer et al. 2016). They found that: Making consumers actively choose to view or skip summary T&C screens increases the likelihood •  they will actually read them and improves repayment performance. • Separating finance charges to make them more salient led to lower default rates. • Segmenting reminder messages by user type improved repayment performance. In Tanzania, CGAP partnered with Busara and Arifu in 2016 to analyze the account activities of over 21,000 farmers, measuring the impact of Arifu’s interactive learning component via SMS on farmers’ behavior (Mazer et al. 2017). They found that: • Interactive SMS increased savings on M-Pawa (doubling, on average). Arifu users took out larger loans, maintained lower outstanding amounts, and repaid faster— •  meaning that customers were more likely to borrow responsibly, improving the sustainability of lending. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 74 Section I Section II Section III Section IV Section V Appendix I Appendix II Role of Consumer In many markets, consumer associations have Representatives in Promoting successfully investigated and advocated for responsible digital credit, helping to identify and Responsible Digital Credit mitigate risks such as predatory lending, misleading terms, and inadequate redress mechanisms. Their Consumer associations are playing an increasingly advocacy has contributed to greater transparency and important role in helping digital borrowers navigate increased accountability in the digital credit ecosystem. the risks associated with digital credit. These risks can undermine trust in DFS and lead to harmful Despite their growing influence, consumer associations outcomes for customers. By offering services such as face persistent challenges. CGAP research highlights debt counseling, financial education, and complaint that limited and unsustainable funding remains a key assistance, consumer associations empower borrowers constraint to scaling their work (Duflos et al. 2021b). In to make more informed decisions and seek redress. In addition, the complexity of DFS and the fragmentation addition, their insights from market research and direct of the regulatory and supervisory landscape can make engagements with digital borrowers can serve as a it challenging for consumer associations to build valuable resource for regulators, and their advocacy in-house capacity to effectively advocate for consumer can push for stronger consumer protection policies. protection regulations. Responsible Digital Credit Solutions by Consumer Representatives Responsible Digital Credit Solutions by Consumer Representatives TABLE 13. Consumer representative solutions by risk management cycle and consumer risk Research Organization Solution Risk Management Cycle Consumer Risk and Issue Prevention Mitigation 1. Consumer advocacy All consumer risks and issues 2. Consumer investigations on recourse Inadequate redress Responsible Digital Credit: Frontier Solutions for Authorities and Providers 75 Section I Section II Section III Section IV Section V Appendix I Appendix II Consumer Advocacy Elevates Consumer Solution 1:  Voices in Financial Policy Making Risk prevention, mitigation • All consumer risks and issues WHY? (THE PA IN P O IN TS ) typically include gathering evidence (e.g., compiling Consumer associations play a critical role in consumer complaints), educating the public via representing the voice of consumers in policy- traditional and social media, and engaging with a broad making processes. Their participation helps to ensure range of stakeholders to advocate for change. that government policies are informed by the lived experiences and needs of consumers. These groups can help to correct power and information asymmetries C H AL L E N GE S Consumer organizations, particularly in EMDEs, face in the financial sector, as highlighted in CGAP’s persistent challenges. Limited funding and resources Elevating the Collective Consumer Voice paper (Duflos constrain their ability to sustain, innovate, or scale et al. 2021b). Their presence in regulatory dialogues their interventions. In addition, many organizations enables vulnerable segments to advocate more require support to build the technical expertise needed effectively for fair treatment and responsible finance. to engage effectively on DFS challenges with policy makers and regulators. WHAT ? (THE SO LUT IO N S ) A consumer advocacy campaign involves a series of actions led by consumer groups to raise awareness and drive change in policy or practice. These campaigns Examples In 2025, the Thailand Consumers Council launched a campaign against preinstalled digital loan apps on smartphone devices that could not be removed. The consumer association has issued several press releases, held conferences, promoted social media messages and videos, gathered consumer stories, demanded concrete actions from authorities and providers, advocated directly with policy makers, and has started a class action lawsuit. As a result, the telco authority prohibited the sales of new Oppo and Realme phone devices. Both companies have indicated that they will send users links with updates to uninstall the loan apps (Thailand Consumers Council 2025). Responsible Digital Credit: Frontier Solutions for Authorities and Providers 76 Section I Section II Section III Section IV Section V Appendix I Appendix II  onsumer Investigations Shed Light Solution 2: C on Inadequate Recourse Risk mitigation • Inadequate redress WHY? (THE PA IN P O IN TS ) collecting real-world data from customers (e.g., through Inadequate redress mechanisms remain a widespread surveys, direct complaints, social media posts, and case issue in digital credit. Consumers often face opaque studies), consumer associations can build an evidence complaint processes, limited dispute resolution base to document consumer harm. This information options, and poor responsiveness from lenders. can be used to run public awareness campaigns, as Many digital credit providers lack clear channels for well as to provide evidence for improving consumer consumers to contest unfair fees, incorrect charges, or protection regulations on digital credit (see Solution 1). fraudulent transactions, disproportionately affecting low-income and digitally inexperienced borrowers. Regulatory gaps and weak enforcement further C H AL L E N GE S AN D L I MI TAT I O N S Consumer associations leading such investigations exacerbate the issue, leaving many borrowers without face numerous challenges, including limited data effective recourse when harmed. access, difficulties in proving systemic issues, mobilizing affected consumers, pushback from WHAT ? (THE SO LUT IO N S ) financial institutions, and resource constraints. Consumer associations can play a critical role by Despite these obstacles, their work is essential in conducting independent investigations into digital exposing unfair practices and advocating for stronger credit products and customers’ experiences. By consumer protections. Examples In 2018, Consumers Korea investigated internet blogs and portals, such as Naver, to uncover flaws in the financial consumer redress system. The investigation revealed that many consumers had not received compensation for mis-sold debt cancellation and debt suspension products added to their credit cards despite regulatory actions (Consumers International n.d.). Their findings highlighted inconsistencies in redress procedures, as some credit card companies issued automatic refunds while others required consumers to file complaints and undergo case-by-case reviews. Based on this evidence, Consumers Korea proposed policy improvements for the Financial Supervisory Service to strengthen consumer protection and redress mechanisms. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 77 Section I Section II Section III Section IV Section V Appendix I Appendix II A Call to Action for Consumer A Call to Action for Funders Representatives and Research Funders have a critical role to play in supporting the Organizations development, testing, and adoption of solutions by research organizations and consumer representatives. Consumer representatives and research organizations They can: are well-positioned to provide key inputs and • Raise awareness among grantees about the incentives that prompt authorities and providers to act importance of addressing consumer risks in digital on responsible digital credit practices. For example, credit through a more holistic and customer-centric they can: lens, one that considers the borrower journey, • Gather more insights from consumers regarding the provider life cycle, and risk management cycle. experiences, vulnerabilities, and outcomes of their • Share promising responsible digital credit solutions personal digital credit journey. from this guide with local research and consumer • Continue to pilot and develop solutions to address organizations and other market actors who can consumer risks in digital credit, especially on topics support or rely on them, and offer technical and with fewer initiatives (e.g., agent, network downtime, financial resources to implement solutions suited to inadequate redress). the local context. • Strengthen institutional capacity to engage in • Support consumer associations to more effectively dialogue and collaborate with authorities and identify and respond to risks in digital credit, providers to develop solutions for preventing, advocate for stronger regulation and provider mitigating, and resolving digital credit consumer practices, help borrowers address risks and enhance risks. their collaboration with government authorities and • Improve understanding of solutions that authorities providers. and providers have adopted in different jurisdictions • Support research and consumer organizations in and identify those better suited to their local piloting solutions that enhance market monitoring context. and gain insights that can inform regulatory or • Inform consumers about prevalent risks associated policy actions and provider standards or practices with digital credit, and ensure access to clear, that address salient and understudied issues (e.g., actionable information on redress mechanisms and behavioral vulnerabilities, agent, network downtime, support channels, especially in the event of disputes fraud). Phone surveys, social media monitoring, or adverse outcomes, including overindebtedness. and mystery shopping provide opportunities for collaboration in market monitoring. • Fund research initiatives to deepen understanding of behavioral vulnerabilities and the root causes of overindebtedness and declining financial health related to digital credit. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 78 Section I Section II Section III Section IV Section V Appendix I Appendix II APPENDIX I Selected Examples of Solutions by Authorities TABLE 14. Examples of authority solutions by phase of risk management cycle and consumer risk Country Example Authority Risk Management Cycle Consumer Risk and Issue Identification Prevention Mitigation Resolution Nigeria Ecosystem Federal All (Cross-agency) Competition Collaboration and Consumer Protection Commission, and others The Data Protection National Privacy Data misuse Philippines Guidelines Commission Kenya Regulation Competition Lack of on Disclosure Authority of transparency of Digital Kenya Transaction Costs European Directives European Behavioral Union addressing Commission vulnerabilities Behavioral Vulnerabilities South Africa Requirements National Credit Unfair treatment on Affordability Regulator Assessments United Regulations Consumer Unfair treatment States addressing Financial Discrimination Protection and Algorithmic Bureau Bias Uganda Guidelines Uganda Over- addressing Microfinance indebtedness Abusive Debt Regulatory Collection Authority Brazil Consumer Ministry of Inadequate Complaints Justice’s redress and Dispute National Resolution Consumer Scheme Secretariat Responsible Digital Credit: Frontier Solutions for Authorities and Providers 79 Section I Section II Section III Section IV Section V Appendix I Appendix II Nigeria: Ecosystem (Cross-Agency) Collaboration4 Federal Competition and Consumer Protection Commission, and others All risk management phases • All consumer risks and issues PROBLEM: part of its broader consumer protection mandate, Nigerian authorities received a growing number of which includes money lending. The Central Bank of complaints about digital lenders engaging in harmful Nigeria participated in the Task Force but did not take debt collection practices, including public shaming on the role of lead regulator and enforcer. and violations of privacy, arbitrary or exploitative interest rates and loan balance calculations, and failure of consumer feedback mechanisms. Initial inquiries R E S U LT S : The registration period ended in mid-2023. As of late revealed that many of the lenders were not legally 2023, there were 164 fully approved digital lenders, established or otherwise licensed to provide the which grew to 320 by January 2025. In addition, the services they were offering. FCCPC has delisted 47 lending apps in collaboration with Google, placed 88 apps and companies on a RESPONSE: watchlist, and placed 42 under conditional approval. In November 2021, the Federal Competition & The FCCPC publishes the full list of digital lending apps Consumer Protection Commission (FCCPC) of Nigeria, and companies in different stages of approval. in partnership with the Independent Corrupt Practices Commission, the National Information Technology Development Agency, and the Central Bank of Nigeria, C H AL L E N GE S : It can be challenging to halt illegal lending apps formed the Joint Regulatory and Enforcement Task housed within bigtech app stores and to ensure the Force to rein in the digital credit apps. In 2022, under firm’s cooperation in imposing delisting. Additionally, the Competition and Consumer Protection Act, FCCPC some Nigerian lenders have resorted to software (acting on behalf of the Task Force) established the workarounds, providing consumers with links to visit Interim Framework and Guidelines for Digital Lending. unregistered websites. During these interactions, These Guidelines set a deadline (that has since been consumers’ protected, private information is illegally extended) for digital money lenders already in business accessed and downloaded. and operating on the Google Play Store to comply with the guidelines or risk removal. R ATI ONALE /C O N T E XT : Given the intersecting risks and abuses extending beyond the central bank’s regulatory perimeter, a coordinated, cross-agency approach was decided upon. FCCPC has jurisdiction over digital lenders as 4 Sources: FCCPC 2021; FCCPC 2023. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 80 Section I Section II Section III Section IV Section V Appendix I Appendix II The Philippines: Data Protection5 General authority: National Privacy Commission Risk prevention • Data misuse PROBLEM: pawnshops, insurance companies, and cooperatives are The National Privacy Commission (NPC) of the exempt from these requirements. Philippines received numerous complaints against lenders using online apps that can be downloaded and installed on mobile phones. The online apps provide R AT I O N AL E /C O N T E X T : The NPC issued the guidelines in response to growing a platform for processing personal data, including concerns over the misuse of personal data by unlicensed the client’s phone contact list, camera, location, and digital lenders. These lenders, which operated beyond storage. The complaints alleged that the lenders the central bank’s regulatory perimeter but within processed their personal data without a lawful basis the purview of NPC, were the subject of numerous and used their personal data and data of people in their complaints related to abusive debt collection. Licensed contact list, causing damage to their reputation, in financial institutions were excluded from the guidelines, violation of their rights and freedoms as data subjects. as they are already supervised by the central bank. The guidelines apply the principle that lenders should be RESPONSE: prevented from collecting any additional personal data The NPC issued Guidelines on the Processing of personal than strictly necessary. data for loan-related transactions through two circulars in 2020 and 2023. These guidelines prohibit lenders operating mobile online applications from accessing or R E S U LT S : The 2023 circular, strengthening the guidelines by harvesting personal information, such as contact lists or adding provisions on just-in-time notices and character social media data, to prevent unfair collection, including references, has been in effect for just over a year. The the harassment and public shaming of borrowers who NPC lists enforcement actions on its site, but other fall behind on payments. Additionally, borrowers’ photos, data on compliance is not available. obtained via access to phone cameras, may only be used for KYC processes. Lenders are required to provide just-in-time notices before obtaining consumer consent C H AL L E N GE S : for data collection, clearly informing borrowers about The rapid emergence of new digital lending apps how their data will be processed. Noncompliance with continues to challenge regulators in the Philippines, these guidelines can result in fines or imprisonment and beyond, to monitor the market carefully and under the Data Privacy Act. The guidelines apply to respond promptly to data abuses. fintech lenders, financing companies, individuals or firms engaged in loan processing, and their third-party service providers. However, specially licensed institutions such as banks, investment houses, savings and loan associations, 5 Source: National Privacy Commission 2020. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 81 Section I Section II Section III Section IV Section V Appendix I Appendix II Kenya: Regulation on Disclosure of Digital Transaction Costs6 General authority: Competition Authority of Kenya Risk prevention • Lack of transparency PROBLEM: uniquely positioned to issue a uniform requirement for Pricing transparency was missing from the market in pre-transaction disclosure across the market. the early 2010s. For instance, Kenya’s mobile money market leader M-Pesa, did not disclose fees on mobile handsets during transactions. The Competition R E S U LT S : A follow-up survey of DFS users in Kenya found that the Authority of Kenya (CAK) found that consumers were percentage of survey participants who could correctly often unaware of the costs of their digital transactions estimate the cost of their last M-Shwari loan (via because providers only revealed the fees after M-Pesa) of KES 200 went up from 52 percent before the consumers had accepted the transaction on their CAK order to 80 percent afterward. mobile devices. This occurred against the backdrop of a highly concentrated and growing mobile money market and a boom in digital credit. C H AL L E N GE S : A central bank directive would have applied only to licensed financial institutions, while the CAK guideline RESPONSE: applied to all DFS providers, including unlicensed digital In 2016, CAK issued an order requiring all DFS providers lenders. However, CAK has more limited supervisory to disclose all applicable charges to customers for and enforcement powers. mobile money transactions (including microloans) prior to the transaction’s completion. This would give consumers an opportunity to cancel, if needed, and to receive a receipt afterward. R ATI ONALE /C O N T E XT : At the time, existing credit regulations required timely disclosure of costs only from licensed financial institutions, not from mobile money operators (MMOs). As a result, the dominant operator opted to disclose full cost information only after the transaction was completed, a practice that was more convenient and profitable. Pressure from the financial authorities and the providers was insufficient to impose a prior disclosure rule. This created significant consumer risks and undercut fair competition by enabling regulatory arbitrage by the mobile money operators. CAK was 6 Sources: Mazer and Garz 2024; Nkhonjera 2017. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 82 Section I Section II Section III Section IV Section V Appendix I Appendix II EU: Behavioral Vulnerabilities7 General authority: European Commission Risk prevention • Behavioral vulnerabilities PROBLEM: R AT I O N AL E /C O N T E X T : The instant, automated, and remote nature of digital In the EU, digitization has significantly transformed the credit can heighten consumer risks. With minimal consumer credit market, introducing practices and risks points of friction in the borrower journey, consumers that were unforeseen under prior Union legislation. In may make hurried, ill-informed and often damaging response, member states adopted diverging provisions, credit choices, which are irreversible. This points to resulting in a fragmented regulatory environment. The a behavioral vulnerability of many consumers, which introduction of updated, harmonized directives on some lenders exploit. consumer credit helps to maintain a high standard of consumer protection across the EU, while maintaining a level playing field in the single EU market. RESPONSE: In 2023, the EU adopted the Consumer Credit Directive (CCD) and the Directive on financial services contracts R E S U LT S : concluded at a distance, which require member states to The directives took effect very recently, and efforts to institute harmonized rules updated to reflect changes in harmonize national laws are underway. the market, especially innovations in digital finance. The CCD requires member states to ensure that C H AL L E N GE S : Low-income and low-financial-literacy consumers in consumers have a right to withdraw from a credit particular are susceptible to accepting digital credit agreement within 14 calendar days without penalty that may turn out to be misleading, damaging, or or obligation to provide justification. This right must fraudulent. Experiments with introducing withdrawal be clearly stated in the credit agreement. Consumers options and other points of friction show that these must notify the lender and repay the principal, and any can be effective in encouraging consumers to pause, interest owed. Additionally, consumers must be entitled reconsider, and look for better options. to early repayment at any time. Under the second directive, the right of withdrawal from distance contracts must be facilitated through a prominent, easy-to- find “withdrawal function” on the provider’s interface. The withdrawal function must be clearly labeled, unambiguous, and comprehensible. A confirmation of withdrawal must also be provided, and easily legible, with an acknowledgement of receipt that is sent to consumers. Member states must also ensure that consumers who have exercised the right of withdrawal are no longer bound by a linked credit agreement. 7 Source: European Union 2023a. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 83 Section I Section II Section III Section IV Section V Appendix I Appendix II South Africa: Product Vulnerabilities8 Financial authority: National Credit Regulator Risk prevention • Unfair treatment PROBLEM: R AT I O N AL E /C O N T E X T : Digital credit models speed up loan sales and Loan sharking and predatory lending proliferated under approvals, which increase the likelihood that consumers apartheid. Since then, South Africa has addressed this may take on new debt without the ability to repay. problem in multiple phases, resulting in the National Credit Act, which established a specialized National Credit Regulator (NCR) under the Department of Trade RESPONSE: and Industry. The Act’s provisions on affordability Under South Africa’s National Credit Act of 2005, apply to all personal credit provision, including the credit providers are required to conduct “affordability fast-growing digital lending sector. assessments” to evaluate a consumer’s financial situation before entering into an agreement. These assessments are intended to help determine if R E S U LT S : borrowers can responsibly take on more debt. Lenders To a large extent, the Act has enhanced financial must consider the following factors: debt repayment inclusion. This conclusion is supported by the statistics history, financial means, prospects and obligations, provided by the National Credit Regulator, which show understanding of risks and costs, and rights and that credit extension increased by 32.49 percent in obligations under the agreement. The assessment must 2017 compared to December 2007, six months after all also consider whether a loan is for business purposes sections of the National Credit Act came into effect. and whether there is a reasonable prospect of success. Extending credit without this assessment is deemed reckless. The provisions on reckless lending and C H AL L E N GE S : Some of the provisions, as applied, appear to hinder affordability assessments apply only to loans granted residential and small business credit, but these have, in to individuals. part, been mitigated by court decisions. 8 Source: Republic of South Africa 2006. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 84 Section I Section II Section III Section IV Section V Appendix I Appendix II US: Discrimination and Algorithmic Bias9 Financial authority: Consumer Financial Protection Bureau Risk prevention • Unfair treatment PROBLEM: R AT I O N AL E /C O N T E X T : There is growing concern about lenders using complex The ECOA’s notice requirements “were designed algorithms to assess creditworthiness while failing to fulfill the twin goals of consumer protection and to provide explanation for any credit denials required education.” These requirements are designed to under the Equal Credit Opportunity Act (ECOA) and advance consumer protection, because “if creditors Regulation B, which implements the Act. Many FSPs know they must explain their decisions . . . they [will] use advanced computational methods as part of effectively be discouraged [from discriminatory their credit decision-making, and the most reputable practices].” The notice requirement also educates typically provide the rationale behind their credit consumers about the reasons for the creditor’s action decisions. Others use “black-box” algorithms, that yield and, therefore, what they need to address in future outputs which cannot be readily explained and may credit applications. reflect biases. R E S U LT S : RESPONSE: The CFPB followed up with a 2023 circular clarifying In a 2022 circular, the CFPB clarified lenders’ duties in that creditors cannot simply use the most analogous this respect. The ECOA requires creditors to provide adverse action rationale listed on the Regulation B applicants with a written notice outlining specific forms if that reason is not accurate and specific under reasons for adverse actions taken against them. the circumstances. The circular also suggests that The specific reasons disclosed must relate to and the CFPB will use the ECOA requirements to increase accurately describe all the key factors considered scrutiny of nontraditional data collected and used in or scored by a creditor. Creditors who use complex AI models. algorithms, including AI or ML, in any aspect of their credit decisions must still provide a notice that discloses the specific principal reasons for C H AL L E N GE S : The opacity of many credit algorithms challenges both taking an adverse action. A creditor cannot justify consumers and regulators. Applying Regulation B is noncompliance based on the mere fact that the complicated by lenders’ reliance on “black-box” models technology it employs to evaluate applications is too and their lack of or inaccurate explanations. complicated or opaque to understand. 9 Source: CFPB 2022. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 85 Section I Section II Section III Section IV Section V Appendix I Appendix II Uganda: Abusive Debt Collection10 Financial authority: Uganda Microfinance Regulatory Authority Risk prevention, mitigation • Overindebtedness PROBLEM: R AT I O N AL E /C O N T E X T : A surge of predatory digital lending caused increasing The rapid growth of digital lending apps and abusive concern, especially given the lack of comprehensive practices necessitated new, bespoke rules for Tier 4 regulation and oversight over more than 2,000 loan apps institutions engaged in digital lending. These entities and 200 bottom-tier licensed Microfinance Institutions. are regulated not by the central bank but by UMRA. While money lenders are often excluded from financial regulation, Uganda’s approach was to bring them under RESPONSE: the 2016 Act along with microcredit providers, thus In 2024, the Uganda Microfinance Regulatory Authority treating their activities on a level playing field. (UMRA) issued a set of new Digital Lending Guidelines for the SACCOS, nondeposit-taking MFIs, and money lenders licensed under the Tier 4 Microfinance R E S U LT S : Institutions and Money Lenders Act (2016). The The impact of these guidelines and the effectiveness of guidelines limit interest penalties and prohibit certain UMRA’s enforcement actions in the market are not clear collection methods. yet, as UMRA’s actions are only beginning to take effect. • Limits on Penalties: No interest penalty on default that exceeds half the initial interest at the time of C H AL L E N GE S : loan offering. The maximum recoverable amount is Regulators worldwide have struggled to keep pace with the sum of principal owed and interest as provided the rapid innovation and proliferation of new digital in the contract, not exceeding the principal due lending apps. This is coupled with the challenge of when the loan becomes nonperforming. avoiding uneven treatment and regulatory arbitrage. • Collection Methods Prohibited: (a) Using threats, obscenity, violence, or other criminal means to physically harm the person, reputation, or property; (b) making unauthorized or unsolicited calls or posting messages to a customer’s contacts who were not party to the loan transaction; and (c) any other conduct that harasses, oppresses, or abuses any person in connection with debt collection. Borrowers must be informed in advance of empaneled or assigned collection agents. 10 Source: Uganda Microfinance Regulatory Authority 2024. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 86 Section I Section II Section III Section IV Section V Appendix I Appendix II Brazil: Consumer Complaints and Dispute Resolution11 General authority: Ministry of Justice’s National Consumer Secretariat Risk resolution • Inadequate redress PROBLEM: R AT I O N AL E /C O N T E X T : Dispute resolution processes in Brazil were located Alternative dispute resolution mechanisms are at the courts and the local offices of the general necessary complements to the legal system, especially consumer protection agency (Procon). With the for individual consumers. Technology has expanded the increase in volume and types of online transactions, ways in which DFS customers can seek out-of-court especially by younger consumers, the types of redress. ODR mechanisms directly connect consumers providers (e.g., fintech lenders), and the relative with FSPs and regulators, increasing convenience, efficiency of online processes, the need for an online, coverage, and speed of dispute resolution for remote alternative dispute resolution mechanism customers at a low cost. became evident. R E S U LT S : RESPONSE: The ODR platform served 1,375 businesses and 5.4 In 2014, Brazil launched a national online dispute million users in 2023. The financial services sector resolution (ODR) platform, consumidor.gov.br, as and electronic payments accounted for 31 percent an informal forum for consumers and businesses of registered complaints, with 12 percent related to to resolve disputes through direct negotiation and credit, debit, and store cards, and 2.7 percent to online private settlement. The Ministry of Justice’s National and mobile payment services. Fintechs and other Consumer Secretariat (Senacon) manages the platform digital lenders have been the subject of thousands but does not intervene in the process. Business of complaints, which have been addressed on the registration on the platform is voluntary, and fintech platform. The Banco Central do Brasil’s conduct lenders have joined. supervision department uses this information to supplement their complaints register for risk analysis. In 2019, “Consumidor.gov.br” was incorporated into the Ministry of Justice platform for electronic judicial process. Upon receiving a consumer complaint, the C H AL L E N GE S : First, the ODR process is automated, with no business has up to 10 days to respond, after which human intervention, so algorithmic bias and other consumers have up to 20 days to accept or reject technological issues may affect it. Second, the the proposed solution. Consumers can also provide platform is used at relatively low rates by individuals a publicly visible business satisfaction rating. If an with lower levels of education, residents of poorer agreement is not reached, consumers can register regions, and elderly populations. Finally, the disparity complaints through traditional redress channels. in power between complainants and FSPs may need Information on disputes, filings, and agreements is greater consumer information and education. publicly available. 11 Sources: Consumidor.gov.br n.d.; Ministry of Justice and Public Security n.d.; Suriani 2022. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 87 Section I Section II Section III Section IV Section V Appendix I Appendix II APPENDIX II Selected Examples of Solutions by Providers TABLE 15. Examples of provider solutions by phase of risk management cycle and consumer risk Provider Type Example Risk Management Cycle Consumer Risk and Issue Identification Prevention Mitigation Resolution Amazon Web Bigtech Enabling Network Services Redundant Failover downtime (AWS) Systems, Disaster Recovery, and Offline Capabilities Microsoft BIgtech Enabling Fair, Unfair treatment Fairlearn Explainable, and Inclusive AI in Digital Credit Google Bigtech Enhancing Lack of Transparency and transparency, Restricting Access Data misuse to Sensitive Data Apple Bigtech Enhancing Lack of Transparency and transparency, Strengthening Data misuse Consumer Protection Responsible Digital Credit: Frontier Solutions for Authorities and Providers 88 Section I Section II Section III Section IV Section V Appendix I Appendix II Amazon Web Services (AWS): Enabling Redundant Failover Systems, Disaster Recovery, and Offline Capabilities Bigtech Risk prevention, mitigation • Network downtime PROBLEM: R E S U LT S : Digital credit platforms depend on always-on AWS enables reliability, scalability, and resilience. infrastructure to deliver time-sensitive services like However, digital lenders should still: loan disbursement and repayment. In low-connectivity • Adopt multicloud or hybrid strategies to mitigate settings, any system failure can disrupt services— provider dependency. undermining user trust, delaying payouts, and • Implement data compliance strategies across increasing reputational risks. Providers like AWS offer regions. cloud-based infrastructure that enables resilient and scalable solutions, minimizing service disruptions. • Carry out transparent downtime communication However, overreliance on centralized systems with borrowers to preserve their trust. introduces operational vulnerabilities. For example, Tala leverages AWS for its cloud infrastructure to maintain platform availability and RESPONSE: service continuity through regional failover and AWS provides a global suite of cloud services that redundancy systems. support: • Redundant Failover Systems: These automatically reroute functions to backup systems during outages C H AL L E N GE S AN D L I MI TAT I O N S : • Complexity and Cost: AWS infrastructure requires to maintain uptime. skilled technical teams and financial planning to • Disaster Recovery: Enables rapid data recovery avoid overspending. by deploying applications and data across multiple • Data Privacy: Compliance with local data privacy availability zones and regions. regulations (e.g., data localization) is critical when • Scalable Cloud Services: Scalable services during operating cross-border.  peak lending periods or repayment cycles, without • Vendor Lock-In: Heavy reliance on a single provider compromising performance. limits flexibility and increases exposure to pricing • Offline Resilience: Enabled cached functions changes or outages. and USSD/SMS–based services for continued • Single Point of Failure: Incidents such as the functionality in low-bandwidth areas. Microsoft and CloudStrike outages demonstrate how centralized platforms can lead to cascading disruptions and service outages.  • Competition Risks: Cloud providers may also enter adjacent financial services, creating potential conflicts of interest. • Security Risks: If sensitive data, IP, and processes are handled by a single provider, they are at risk if not properly managed. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 89 Section I Section II Section III Section IV Section V Appendix I Appendix II Microsoft Fairlearn: Enabling Fair, Explainable, and Inclusive AI in Digital Credit Bigtech All risk management phases • Unfair treatment PROBLEM: R E S U LT S : AI-based credit scoring can unintentionally produce By using Fairlearn, digital lenders can: biased or discriminatory outcomes if trained on • Detect and reduce bias in credit decisioning.   nonrepresentative data such as overweighted income, • Improve transparency for regulators and users. gender, or geographic variables. This can lead to unfair loan rejections and pricing disparities, especially in • Support regulatory compliance with emerging AI EMDEs with diverse populations, insufficient data, and fairness and discrimination laws.  limited regulatory oversight. • Make more accurate and fair credit decisions.  For example, digital lenders such as Tala, Branch, RESPONSE: Lenddo, and CreditMantri have used Fairlearn to Microsoft’s Fairlearn is a community-driven, improve their processes. open-source Python library designed to improve the fairness and transparency of ML models. It helps developers evaluate and mitigate bias in credit scoring, C H AL L E N GE S AN D L I MI TAT I O N S : as well as comply with fairness-related AI principles. • Data Quality and Availability: To effectively Key features include: mitigate bias, it would require large, diverse, and • Fairness metrics: Demographic parity to measure high-quality data sets. equal prediction rates across groups, equalized • Scoring Model Complexity: Fairlearn is better odds to measure equal true positive and false suited to interpretable models; integration with positive rates across groups, and disparate metrics proprietary “black-box” scoring systems can be to measure the ratio of positive outcomes between challenging. groups. • Fairness Vs. Accuracy Trade-Offs: Optimizing for • Bias detection and mitigation algorithms. fairness may slightly reduce predictive accuracy— • Tools for model transparency and interpretability. requiring careful calibration. • Fairness and Antidiscrimination: It is also worth When conducting a fairness assessment, three main noting that addressing algorithm fairness is different steps should be included: (i) identify who will be from complying with antidiscrimination laws. harmed, (ii) identify the types of harms anticipated, • Ongoing Maintenance: Fairness must be continually and (iii) define fairness metrics based on the monitored as data and user populations evolve over anticipated harms. time. Responsible Digital Credit: Frontier Solutions for Authorities and Providers 90 Section I Section II Section III Section IV Section V Appendix I Appendix II Google: Enhancing Transparency and Restricting Access to Sensitive Data Bigtech All risk management phases • Lack of transparency, data misuse PROBLEM: R E S U LT S : Digital personal lending can operate with limited • Increased borrower protection by limiting oversight in some cases, leading to data misuse, a exploitative practices and ensuring transparent lack of transparency of loan terms, and predatory lending terms. lending practices. Inconsistent disclosure of terms and • Enhanced data privacy by restricting unnecessary unrestricted access to customers’ sensitive data— access to sensitive user information. such as contacts, photos, location, and messages— • Better regulatory alignment with global standards undermine consumer protection, particularly in EMDEs. on financial consumer protection and data privacy. RESPONSE: For example, Google collaborates with regulators and Google has strengthened borrower protection by industry associations to curb predatory lending apps enhancing transparency, safeguarding data, and in its store. For instance, to be listed on the Google enforcing fair lending practices: Play Store in emerging markets such as India, Kenya, Nigeria, and the Philippines, digital lenders must prove • Google Play Policy Update requires all personal compliance with local licensing and regulations. loan apps listed in the Google Play Store to clearly disclose the APR, repayment period, and loan terms (Google’s Digital Lending Policy Updates in 2021 C H AL L E N GE S AN D L I MI TAT I O N S : and 2023). • Limited Enforcement Scope: Policies apply only to • Apps offering loans of 60 days or less or with APRs lending apps distributed via the official Google Play exceeding 36 percent are banned in the US and Store. Digital apps installed through sideloading or restricted in parts of Asia and Africa. third-party stores may still circumvent protections. • Implementation Gaps Across Regions: Digital lending apps are prohibited from accessing Enforcement varies geographically, with some sensitive customer data such as photos, contacts, emerging markets facing slower rollout or videos, precise location, texts, and call logs, unless inconsistent enforcement. necessary for loan underwriting and service delivery. • Developer Circumvention Risks: Some lenders may (Google’s Data Safety Section requires apps to disclose attempt to rebrand or repackage their lending apps data collection and sharing practices, enhancing user to evade scrutiny, requiring constant monitoring. awareness and data privacy compliance.) Responsible Digital Credit: Frontier Solutions for Authorities and Providers 91 Section I Section II Section III Section IV Section V Appendix I Appendix II Apple: Enhancing Transparency and Strengthening Consumer Protections Bigtech All risk management phases • Lack of transparency, data misuse PROBLEM: R E S U LT S : Predatory digital lending practices, including high-cost • Increased borrower transparency by requiring short-term loans and misuse of consumer data, disclosure of clear, consistent loan terms. have raised significant concerns globally. Borrowers, • Strengthened data privacy by limiting unnecessary particularly in EMDEs, are vulnerable to nontransparent access to sensitive personal information. loan terms, unfair pricing, and privacy violations. • Reduction in predatory lending by imposing stricter Inconsistent standards across app stores further app vetting and enforcement on the App Store. enabled some lenders to exploit consumers through hidden fees, aggressive data collection, and unclear • Alignment with global standards on financial repayment obligations. consumer protection and data rights. For example, Apple’s updated App Store policies RESPONSE: have helped regulators in emerging markets, such Apple has strengthened borrower protections through as India and parts of Southeast Asia, identify and updated App Store policies for personal loan apps by: restrict predatory loan apps. Lenders must meet local • Mandatory Disclosure of Key Loan Terms: App regulatory compliance requirements to remain listed. Store guidelines require apps offering personal loans to clearly disclose APR, loan repayment terms, and due dates upfront. C H AL L E N GE S AN D L I MI TAT I O N S : • Limited Jurisdiction: Rules only apply to apps • Banning Short-Term, High-Cost Loans: Apps distributed via the official Apple App Store; apps offering loans of 60 days or less or with APRs downloaded via other channels may still bypass exceeding 36 percent are banned from the App protections. Store in several regions. • Enforcement Gaps Across Regions: • Enhanced User Privacy: Apps are restricted from Implementation consistency varies, with emerging harvesting sensitive consumer data—such as markets sometimes facing slower enforcement. contacts, photos, call logs, or device information— • Developer Circumvention Risks: Some lenders may unless it is essential for loan processing. attempt to rebrand or repackage their lending apps • Strengthened Enforcement: Apple enforces to evade scrutiny, requiring constant monitoring. compliance through periodic app reviews and updates. (Apple revised its App Store Guidelines in 2021 and 2022.) Responsible Digital Credit: Frontier Solutions for Authorities and Providers 92 References Accenture. 2023. “Key Disruptions Shaping the Future of Australian Government. 2024. “National Consumer Credit Lending.” https://bankingblog.accenture.com/key-disruptions- Protection Act 2009.” Compilation, December 11. https://www. shaping-future-lending legislation.gov.au/C2009A00134/latest/text Agarwal, Poonam, Nupur Sonar, and Stephanie Hegarty. 2023. Australasian Legal Information Institute (AustLII). 2010. “Inside the deadly instant loan app scam that blackmails with “NATIONAL CONSUMER CREDIT PROTECTION REGULATIONS nudes.” BBC, October 10. https://www.bbc.com/news/world- 2010.” https://www8.austlii.edu.au/cgi-bin/viewdoc/au/legis/ asia-india-66964510 cth/consol_reg/nccpr2010486/index.html Alliance for Financial Inclusion (AFI). 2020. “Policy Framework on ———. 2018. “National Credit Laws.” https://austlii. Responsible Digital Credit.” February. https://www.afi-global. community/wiki/ACTLawHbk/NationalCreditLaws org/sites/default/files/publications/2020-04/EN_Policy_ Framework_for_Responsible_Digital_Credit.pdf Banco Central do Brasil. 2023. “BC edita regulamentação para a execução do compartilhamento de informações sobre indícios Allied Market Research. 2021. “FinTech Lending Market Size, de fraudes.” https://www.bcb.gov.br/detalhenoticia/17996/nota Share, Competitive Landscape and Trend Analysis Report, by Offering, Business Model, Enterprise Size and Lending Banco de Portugal. 2020. “Circular Letter No Channel: Global Opportunity Analysis and Industry Forecast, CC/2020/00000044. Best practices applicable to the selling of 2021–2030.” October. https://www.alliedmarketresearch.com/ retail banking products and services through digital channels.” fintech-lending-market-A14263 https://clientebancario.bportugal.pt/sites/default/files/ relacionados/noticia/CircularLetter44_2020.pdf Alushula, Patrick. 2020. “Digital Lenders Cut Credit to Kenyans After CBK Directive.” The Standard, September BankBuddy. 2024. “Chatbot for Personal Loan.” https:// 12. https://www.standardmedia.co.ke/business/news/ bankbuddy.ai/chatbot-for-personal-loan/ article/2001430705/digital-lenders-cut-credit-to-kenyans- Bank Negara Malaysia. 2025. “Annual Report 2024.” https:// after-cbk-directive www.bnm.gov.my/documents/20124/17493532/ar2024_en_ Alvarez Ruiz, Lucciana. 2023. “Gender Bias in AI: An book.pdf/ Experiment with ChatGPT in Financial Inclusion.” Center Bartholomae, Suzanne, and Jonathan Fox. 2021. “A Decade for Financial Inclusion (blog), August. https://www. Review of Research on College Student Financial Behavior centerforfinancialinclusion.org/gender-bias-in-ai-an- and Well-Being.” Journal of Family and Economic Issues 42 (5). experiment-with-chatgpt-in-financial-inclusion/ https://www.researchgate.net/publication/350385214_A_ Amazon Web Services (AWS). 2024. “Data Encryption in Decade_Review_of_Research_on_College_Student_Financial_ Amazon Bedrock.” https://docs.aws.amazon.com/bedrock/ Behavior_and_Well-Being latest/userguide/data-encryption.html Basel Committee on Banking Supervision. 2016. “Guidance on Apple. 2024. “App Store Review Guidelines.” https://developer. the Application of the Core Principles for Effective Banking apple.com/app-store/review/guidelines/ Supervision to the Regulation and Supervision of Institutions Relevant to Financial Inclusion.” Geneva: Bank for International Atlan. 2023. “Data Ethics 101.” https://atlan.com/ Settlements. https://www.bis.org/bcbs/publ/d383.htm data-ethics-101/ Responsible Digital Credit: Frontier Solutions for Authorities and Providers 93 Bento, José et al. 2022. “TimeSHAP: Explaining Recurrent Boeddu, Gian, Jennifer Chien, Ivor Istuk, and Ros Grady. 2021. Models Through Sequence Perturbations.” Feedzai Research. “Consumer Risks in Fintech: New Manifestations of Consumer https://research.feedzai.com/oshuzuxa/2022/08/Bento_ Risks and Emerging Regulatory Approaches.” Policy Research TimeSHAP_KDD2021.pdf Paper, World Bank Group (WBG) and Ministry of Foreign Affairs, The Netherlands. https://documents1.worldbank.org/curated/ Better Than Cash Alliance (BTCA). 2021. “UN Principles for en/515771621921739154/pdf/Consumer-Risks-in-Fintech- Responsible Digital Payments.” https://responsiblepayments. New-Manifestations-of-Consumer-Risks-and-Emerging- org/ Regulatory-Approaches-Policy-Research-Paper.pdf Biometric Update. 2018. “BioCatch Signs Up Seven Tier Branch. 2024. “Fraud Analytics.” https://help.branch.io/v1/docs/ One Banks in Latin America for Behavioral Biometric Fraud fraud-analytics Prevention.” https://www.biometricupdate.com/201810/ biocatch-signs-up-seven-tier-one-banks-in-latin-america-for- ———. N.d. “How it Works.” https://branch.co/faq behavioral-biometric-fraud-prevention Burlando, Alfredo, Michael A. Kuhn, and Silvia Prina. 2023. “Too ———. 2020. “South African Organizations Implement Voice Fast, Too Furious? Digital Credit Delivery Speed and Repayment and Face Biometrics to Detect, Reduce Fraud.” https://www. Rates.” Working Paper, May 17. https://static1.squarespace.com/ biometricupdate.com/202004/south-african-organizations- static/57869cbb2e69cf3d5af77419/t/64b6008741fcb13f8a75 implement-voice-and-face-biometrics-to-detect-reduce- 70d1/1689649290419/BKP_2023.05.17.pdf fraud Burman, Anirudh. 2023. “Understanding India’s New Data ———. 2021a. “Biometrics Adoption in Latin America Driving Protection Law.” Working Paper, Carnegie India, October. Innovatrics’ Market Numbers.” https://www.biometricupdate. https://carnegieendowment.org/research/2023/10/ com/202103/biometrics-adoption-in-latin-america-driving- understanding-indias-new-data-protection-law?lang=en. innovatrics-market-numbers Business & Human Rights Resource Centre. 2021. “Kenya: ———. 2021b. “AWS, Google Cloud, IBM, and Microsoft Each Safaricom to Hide Identity of Users of Mobile Money Platform Have Strengths in the Public Cloud, According to Wall Street.” Following Reports of Widespread Misuse of Clients’ Data by https://www.businessinsider.com/aws-google-cloud-ibm- Firms.” https://www.business-humanrights.org/en/latest-news/ microsoft-public-cloud-strengths-wall-street-2021-6 kenya-safaricom-to-hide-identity-of-users-of-mobile-money- platform-following-reports-of-widespread-misuse-of-clients- ———. 2021c. “TOC (Acquired by Sovos).” https://www. data-by-firms/ biometricupdate.com/companies/toc Caire, Dean, and Maria Fernandez Vidal. 2024. “Gender- ———. 2024a. “India’s Central Bank Launches Lending Intentional Credit Scoring.” Technical Guide, Washington, D.C.: Platform Connected to Aadhaar KYC.” https://www. CGAP. https://www.cgap.org/research/publication/gender- biometricupdate.com/202408/indias-central-bank-launches- intentional-credit-scoring lending-platform-connected-to-aadhaar-kyc Campbell Attorneys. N.d. “Consumer Law, Credit Law and ———. 2024b. “Voice Biometrics – Service Directory.” Insolvency Law for Debt Help and Credit Restoration.” https:// https://www.biometricupdate.com/service-directory/voice- legalrights.co.za/ biometrics Cash App. 2024. “Cash App.” https://cash.app/ ———. 2025. “Mastercard Integrates Feedzai Behavioral Biometrics for Fraud Protection.” https://www.biometricupdate. Center for Effective Global Action (CEGA). 2023. “Gender- com/202502/mastercard-integrates-feedzai-behavioral- Differentiated Credit Algorithms Using Machine Learning.” biometrics-for-fraud-protection Research Brief, Berkeley, CA: Center for Effective Global Action. https://cega.berkeley.edu/research/gender-differentiated- Björkegren, Daniel, Joshua Blumenstock, Omowunmi Folajimi- credit-algorithms-using-machine-learning/. Senjobi, Jacqueline Mauro, and Suraj R. Nair. 2022. “Instant Loans Can Lift Subjective Well-Being: A Randomized Evaluation ———. N.d. “Digital Credit Observatory.” Center for Effective of Digital Credit in Nigeria.” Working Paper. https://arxiv.org/ Global Action, Berkeley, CA. https://cega.berkeley.edu/ abs/2202.13540 collection/digital-credit-observatory/? Responsible Digital Credit: Frontier Solutions for Authorities and Providers 94 Center for Effective Global Action (CEGA) and Innovations Cisco. 2021. “Meeting Financial Regulatory Requirements with for Poverty Action (IPA). 2024. “Mobile Instant Credit: Impacts, Duo Multi-Factor Authentication.” https://blogs.cisco.com/ Challenges, and Lessons for Consumer Protection.” Report, financialservices/meeting-financial-regulatory-requirements- Berkeley, CA: CEGA. https://reports-cega.berkeley.edu/mobile- with-duo-multi-factor-authentication instant-credit-report/ Citizens Advice. 2019. “Payday Loan Report 2.” London: Central Bank of Kenya. 2022. “The Central Bank of Kenya Citizens Advice. https://www.citizensadvice.org.uk/Global/ Act.” Kenya Gazette Supplement, March 18. https://www. CitizensAdvice/Debt%20and%20Money%20Publications/ centralbank.go.ke/wp-content/uploads/2022/03/L-.N.- Payday%20Loan%20Report%202.pdf. No.-46-Central-Bank-of-Kenya-Digital-Credit-Providers- Regulations-2022.pdf Consumer Credit Oversight Board Task Force. N.d. “About Us.” Kuala Lumpur. https://ccob.my/about/. Cerise+Social Performance Task Force (SPTF). 2024. “DFS Standards: Management Standards for the Responsible Consumer Financial Protection Bureau (CFPB). 2022a. “CFPB Provision of Digital Financial Services.” https://cerise-sptf.org/ Acts to Protect the Public from Black-Box Credit Models Using digitalfinancial-services/ Complex Algorithms.” https://www.consumerfinance.gov/ about-us/newsroom/cfpb-acts-to-protect-the-public-from- CGAP. 2017. “Emerging Landscape of Digital Credit.” Reading black-box-credit-models-using-complex-algorithms/? Deck, Washington, D.C.: CGAP. https://www.cgap.org/research/ reading-deck/emerging-landscape-of-digital-credit. ———. 2022b. “Consumer Financial Protection Circular 2022-03.” Circular, Washington, D.C, May 26.  https://www. ———. 2019. “Digital Credit Models for Small Businesses.” consumerfinance.gov/compliance/circulars/circular-2022- Publication, Washington, D.C.: CGAP. https://www.cgap. 03-adverse-action-notification-requirements-in-connection- org/research/publication/digital-credit-models-for-small- with-credit-decisions-based-on-complex-algorithms/ businesses ———. 2023. “Civil Penalty Fund.” https://www. ———. 2022a. “Central Bank of Kenya, CGAP, FSD Kenya, consumerfinance.gov/enforcement/payments-harmed- Kenya National Bureau of Statistics.” Publication, Washington, consumers/civil-penalty-fund/. D.C.: CGAP. https://www.cgap.org/research/publication/ market-monitoring-kenya-country-case Consumers International. N.d. “Maximising consumer voices in digital financial services policy-making: Consumer ———. 2022b. “Bank of Tanzania, CGAP, FSDA, FSDT.” organisations’ engagement strategies.” https://www. Publication, Washington, D.C.: CGAP. https://www.cgap.org/ consumersinternational.org/media/552336/maximising- research/publication/market-monitoring-tanzania-country- consumer-voices-in-digital-finance-report.pdf case Consumidor.gov.br. N.d. “ATENÇÃO, CONSUMIDOR!” https:// ———. 2025a. “Data-Driven Financial Services.” https://www. www.consumidor.gov.br/pages/principal/?1745962944410 cgap.org/topics/collections/data-driven-financial-services Cook, Tamara, and Claudia McKay. 2015. “How M-Shwari Works: ———. 2025b. “Digital Credit.” https://www.cgap.org/topics/ The Story So Far.” Forum 10,. Washington, D.C.: CGAP and FSD collections/digital-credit Kenya. https://www.cgap.org/research/publication/how-m- shwari-works-story-so-far   CGAP, BIS, IMF, UNSGSA, and World Bank. 2024. “Key Considerations for Open Finance.” Washington, D.C.: CGAP. Cornelli, Giulio, Jon Frost, Leonardo Gambacorta, Raghavendra https://www.cgap.org/research/key-considerations-for- Rau, Robert Wardrop, and Tania Ziegler. 2020. “Fintech and openfinance Big Tech Credit: A New Database.” BIS Working Paper No. 887, Bank for International Settlements, Monetary and Economic Chalwe-Mulenga, Majorie, Eric Duflos, and Gerhard Coetzee. Department, September. https://www.bis.org/publ/work887. 2022. “The Evolution of the Nature and Scale of DFS Consumer pdf Risks: A Review of Evidence.” Slide Deck, Washington, D.C.: CGAP. https://www.cgap.org/research/reading-deck/ D91 Labs. N.d. “Playbook on Designing Financial Products evolution-of-nature-and-scale-of-dfs-consumer-risks-review- for Women.” D91 Labs. https://d91labs.github.io/Gender- of-evidence Intelligent-Design-Playbook.pdf Responsible Digital Credit: Frontier Solutions for Authorities and Providers 95 Deloitte. 2023. “Data Privacy and Security: The Connectivity and Dvara Research and CGAP. 2021. “Workshop Proceedings: A Mobile Trends Survey 2023.” https://www2.deloitte.com/us/en/ Convening on ‘Emerging Customer Risks in Digital Lending insights/industry/telecommunications/connectivity-mobile- in India.’ ” August. https://dvararesearch.com/wp-content/ trends-survey/2023/data-privacy-and-security.html uploads/2024/01/A-Convening-on-Emerging-Customer- Risks-in-Digital-Lending-in-India.pdf Demircan, Muhammed. 2023. “Deployers of High-Risk AI Systems: What Will Be Your Obligations Under the EU AI Act?” Economic Times. 2024. “RBI Issues New Digital Lending June 2. https://competitionlawblog.kluwercompetitionlaw. Guidelines for Banks & Lenders to Protect Borrowers.” February com/2023/06/02/deployers-of-high-risk-ai-systems-what- 23. https://economictimes.indiatimes.com/wealth/borrow/ will-be-your-obligations-under-the-eu-ai-act/ rbi-issues-new-digital-lending-guidelines-for-banks-lenders- to-protect-borrowers/articleshow/939%E2%80%A6 Digital Financial Services Association of Kenya (DFSAK). 2019. “Code of Conduct for Responsible Lending for the Digital Entrust. 2024. “KeyControl – Key Management Products.” Lending Institutions.” https://www.dfsak.co.ke/assets/files/ https://www.entrust.com/products/key-management/ dlak_code_of_conduct.pdf keycontrol Duflos, Eric, and Corinne Riquet. 2024. “Surveys Reveal a Path to European Banking Authority. 2023. “The EBA’s recent mystery Responsible Digital Finance in WAEMU.” Blog, Washington, D.C.: shopping exercise shows the value this tool adds to the CGAP. https://www.cgap.org/blog/surveys-reveal-path-to- supervisory tasks of national authorities.” Press Release. https:// responsible-digital-finance-in-waemu www.eba.europa.eu/publications-and-media/press-releases/ ebas-recent-mystery-shopping-exercise-shows-value-tool- Duflos, Eric, Daryl Collins, Jayshree Venkatesan, and Juan Carlos adds-supervisory-tasks-national-authorities Izaguirre. 2021a. “Analyzing Social Media to Spot Digital Credit Consumer Risks in India.” Blog, Washington, D.C.: CGAP. https:// European Central Bank. 2018. “Guide to assessments of www.cgap.org/blog/analyzing-social-media-to-spot-digital- fintech credit institution licence applications.” https://www. consumer-credit-risks-in-india bankingsupervision.europa.eu/ecb/pub/pdf/ssm.201803_ guide_assessment_fintech_credit_inst_licensing.en.pdf Duflos, Eric, Mary Griffin, and Myra Valenzuela. 2021b. “Elevating the Collective Consumer Voice in Financial Regulation.” European Parliament. 2024. “Artificial Intelligence Act.” March Working Paper, Washington, D.C.: CGAP. https://www.cgap.org/ 13. Strasbourg. https://www.europarl.europa.eu/doceo/ research/publication/elevating-collective-consumer-voice-in- document/TA-9-2024-0138_EN.html? financial-regulation European Union. 2016. “Regulation (EU) 2016/679 of the Duflos, Eric, Juan Carlos Izaguirre, Leena Datwani, Abhishek European Parliament and of the Council of 27 April 2016 on the Mishra, Daryl Collins, Pravarakya Reddy Bathula, Leona Mathias, protection of natural persons with regard to the processing and Dhanashri Sawant. 2023. “Social Media Monitoring to of personal data and on the free movement of such data, Assess Consumer Risks in Digital Credit Apps: Guidance for and repealing Directive 95/46/EC (General Data Protection Supervisors from an India Pilot.” Reading Deck, Washington, Regulation).” Official Journal of the European Union, May 4. D.C.: CGAP. https://www.cgap.org/research/reading-deck/ https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng social-media-monitoring-to-assess-consumer-risks-in-digital- credit-apps ———. 2023a. “Directive (EU) 2023/2225 of the European Parliament and of the Council of 18 October 2023 on credit Duflos, Eric, Juan Carlos Izaguirre, Sai Krishna Kumaraswamy, agreements for consumers and repealing Directive 2008/48/ Laura Brix Newbury, and Myra Valenzuela. 2024. “Responsible EC.” Official Journal of the European Union, October 30. https:// Digital Finance Ecosystem (RDFE): A Conceptual Framework.” eur-lex.europa.eu/eli/dir/2023/2225/oj Working Paper, Washington, D.C.: CGAP. https://www.cgap.org/ research/publication/responsible-digital-finance-ecosystem- ———. 2023b. “Directive (EU) 2023/2673 of the European rdfe-conceptual-framework Parliament and of the Council of 22 November 2023 amending Directive 2011/83/EU as regards financial services contracts Dvara Research. 2023. “Financial Customer Protection: An concluded at a distance and repealing Directive 2002/65/EC Interim Research Report.” August. https://dvararesearch. (Text with EEA relevance).” Official Journal of the European com/centre-for-customer-protection/wp-content/ Union, November 11. https://eur-lex.europa.eu/legal-content/ uploads/2023/08/Dvara_Research_30082023_V19.pdf EN/TXT/PDF/?uri=OJ%3AL_202302673 Responsible Digital Credit: Frontier Solutions for Authorities and Providers 96 EY. 2021. “Assessing and Mitigating Unfairness in Credit Models.” Fernandez Vidal, Maria, and Arisha Salman. 2023. “Global https://www.ey.com/en_ca/insights/financial-services/ Landscape: Data Trails of Digitally Included Poor (DIP) People.” assessing-and-mitigating-unfairness-in-credit-models Reading Deck, Washington, D.C.: CGAP. https://www.cgap. org/research/reading-deck/global-landscape-data-trails-of- Federal Competition and Consumer Protection Commission digitally-included-poor-dip-people (FCCPC). 2021. “FCCPC, ICPC, EFCC, NITDA, NHRC AND CBN to Jointly Investigate Rights Violations in Money Lending Financial Conduct Authority (FCA). 2022. “FG22/5 Final Industry.” Nigeria. https://fccpc.gov.ng/fccpc-icpc-efcc-nitda- non-Handbook Guidance for firms on the Consumer Duty.” nhrc-and-cbn-to-jointly-investigate-rights-violations-in- Finalised Guidance. https://www.fca.org.uk/publication/ money-lending-industry/ finalised-guidance/fg22-5.pdf ———. 2022a. “FCCPC, ICPC, EFCC, NITDA, NHRC AND FinCoNet. 2017. “Report on the Digitalisation of Short-Term, CBN to Jointly Investigate Rights Violations in Money Lending High-Cost Consumer Credit.” November. https://www.finconet. Industry: Initial Enforcement Operations by Joint Task Force org/finconetreportonthedigitalisationofshort-termhigh- (FCCPC, NITDA and ICPC).” Nigeria. https://fccpc.gov.ng/ costconsumercredit.htm fccpc-icpc-efcc-nitda-nhrc-and-cbn-to-jointly-investigate- rights-violations-in-money-lending-industry-initial- Finovate. 2024. “Feedzai Archives.” https://finovate.com/ enforcement-operations-by-joint-task-force-fccpc-nitda-and- category/feedzai/ icpc/ Fintech Association for Consumer Empowerment (FACE). N.d. ———. 2022b. “Limited Interim Regulatory/ Registration FACE Homepage. https://faceofindia.org/ Framework And Guidelines For Digital Lending, 2022.” Nigeria. Forbes. 2016. “How Tala Mobile Is Using Phone Data to https://fccpc.gov.ng/wp-content/uploads/2023/10/LIMITED- Revolutionize Microfinance.” https://www.forbes.com/sites/ INTERIM-REGULATORY_-REGISTRATION-FRAMEWORK-FOR- forbestreptalks/2016/08/29/how-tala-mobile-is-using-phone- DIGITAL-LENDING-2022.pdf data-to-revolutionize-microfinance/ ———. 2023a. “FCCPC and Partners Investigate Rights Francis, Eilin, Joshua Blumenstock, and Jonathan Robinson. Violations in Money Lending Industry.” Nigeria. https://fccpc. 2019. “Digital Credit: A Snapshot of the Current Landscape gov.ng/fccpc-icpc-efcc-nitda-nhrc-and-cbn-to-jointly- and Open Research Questions.” E-Scholarship Paper, investigate-rights-violations-in-money-lending-industry/ Berkeley, CA: Center for Effective Global Action. https:// ———. 2023b. “Further Updates on Investigation of Rights escholarship.org/content/qt88r1j7sz/qt88r1j7sz_ Violations in Money Lending.” https://fccpc.gov.ng/further- noSplash_886aa36c72ccf246b519cc9213be19ca. update-further-and-continuing-investigation-of-rights- pdf?t=osxr66 violations-in-money-lending-industry-and-release-of-interim- Fraud.com. 2024. “Artificial Intelligence: A Double-Edged Sword regulatory-framework/ for Fraud Detection.” https://www.fraud.com/post/artificial- Feedzai. 2024a. “Cut Fraud Approval Time: Banco BV Partners’ intelligence Success with Feedzai.” https://www.feedzai.com/resource/ Freshworks. 2024. “Customer Success Story: M-KOPA.” https:// cut-fraud-approval-time-banco-bv-partners-success-with- www.freshworks.com/customers/stories/m-kopa/ feedzai Fu, Jonathan, and Mrinal Mishra. 2022. “Combatting Fraudulent ———. 2024b. “Feedzai Launches ScamAlert GenAI Fraud and Predatory Fintech Apps with Machine Learning.” Working Agent.” https://www.feedzai.com/pressrelease/feedzai- Paper. https://poverty-action.org/sites/default/files/2023-05/ launches-scamalert-genai-fraud-agent/ Combatting-Fraudulent-and-Predatory-Fintech-Apps-with- Fernandez Vidal, Maria, and Fernando Barbon. 2019. “Credit Machine-Learning-Policy-Brief-Fu-Mishra-Feb-2022-Project- Scoring in Financial Inclusion.” CGAP Technical Guide, Page.pdf Washington, D.C.: CGAP. https://www.cgap.org/research/ Funa, Dennis B. 2017. “Financial Sector Forum.” Opinion, publication/credit-scoring-in-financial-inclusion Insurance Commission, April 26. https://www.insurance.gov.ph/ financial-sector-forum/ Responsible Digital Credit: Frontier Solutions for Authorities and Providers 97 G20/OECD Task Force on Financial Consumer Protection. 2021. ———. 2024. “Consumer Protection in respect of Use of “Compendium of Effective Approaches for Financial Consumer Generative Artificial Intelligence.” Circular. August 19. https:// Protection in the Digital Age.” https://www.oecd.org/content/ brdr.hkma.gov.hk/eng/doc-ldg/docId/20241107-1-EN dam/oecd/en/topics/policy-sub-issues/financial-consumer- protection/Effective-Approaches-FCP-Principles_Digital_ Hwang, Byoung-Hwa, and Camilo Tellez. 2016. “The Proliferation Environment.pdf of Digital Credit Deployments.” Brief. Washington, D.C.: CGAP. https://www.cgap.org/research/publication/proliferation-of- Gambacorta, Leonardo, Yiping Huang, Han Qiu, and Jingyi digital-credit-deployments Wang. 2019. “How Do Machine Learning and Non-Traditional Data Affect Credit Scoring?” BIS Working Papers No. 834. IBM. 2024. “Cost of a Data Breach Report 2024.” https://www. https://www.bis.org/publ/work834.pdf ibm.com/downloads/documents/us-en/107a02e94948f4ec GeeksforGeeks. 2024. “AWS vs Google Cloud Platform vs India Today. 2024. “Google Deletes 2,200 Fake Loan Apps from Azure.” https://www.geeksforgeeks.org/aws-vs-google-cloud- Play Store: Here is How to Be Safe.” February 6. https://www. platform-vs-azure/ indiatoday.in/technology/news/story/google-deletes2200- fake-loan-apps-from-play-store-here-is-how-to-be- Gerhardt, M. 2009. “Consumer Bankruptcy Regimes and safe-2498426-2024-02-06 Credit Default in the US and Europe: A Comparative Study.” CEPS Working Document No. 318/July. https://aei.pitt. Innovations for Poverty Action (IPA). 2021. “UCC-IPA-MNO edu/11336/1/1887.pdf Complaints Data Analysis Final Report.” Washington, D.C.: Innovations for Poverty Action, September. https://poverty- Google. 2024a. “Google Account Dashboard.” https:// action.org/sites/default/files/publications/UCC-IPA-MNO- myaccount.google.com/intro/dashboard?hl=en-US Complaints-Data-Analysis-Final-Report-September-2021- FINAL.pdf ———. 2024b. “Personal Loan App Requirements for Google Play.” https://support.google.com/googleplay/android- ———. 2022a. “Using Consumer Complaint Data for Targeted developer/answer/9876821 Consumer Protection.” Blog Post, Washington, D.C.: Innovations for Poverty Action. https://poverty-action.org/blog/using- Google Cloud. 2023. “Navigating RBI’s Digital Lending consumer-complaint-data-targeted-consumer-protection Guidelines: A Cloud Perspective.” https://services.google.com/ fh/files/misc/google_cloud_rbi_digital_lending_guidelines_ ———. 2022b. “Translating Digital Credit Transaction Data whitepaper_2023.pdf for Consumer Protection Supervision.” Research Report, Washington, D.C.: Innovations for Poverty Action. https:// Government of Canada. 2025. “Canadian Anti-Fraud Centre.” poverty-action.org/translating-digital-credit-transaction-data- https://antifraudcentre-centreantifraude.ca/index-eng.htm consumer-protection-supervision Hamid, Fazelina S. 2025. “Behavioral Biases ———. 2023a. “Consumer Protection in Digital Finance and Over-Indebtedness in Consumer Credit: Surveys.” Survey Report, Washington, D.C.: Innovations Evidence from Malaysia.” Cogent Economics & for Poverty Action. https://poverty-action.org/consumer- Finance 13 (1). https://www.tandfonline.com/doi/ protection-digital-finance-surveys full/10.1080/23322039.2024.2449191?src=&#d1e134. ———. 2023b. “In the Philippines, Chatbots Help Consumer Hong Kong Monetary Authority (HKMA). 2022. “Enhancing Voices Be Heard by Financial Institutions.” Blog Post, Consumer Protection in Respect of ‘Buy Now, Pay Later’ Washington, D.C.: Innovations for Poverty Action. https:// Products.” Circular, September 2. https://brdr.hkma.gov.hk/eng/ poverty-action.org/blog/philippines-chatbots-help-consumer- doc-ldg/docId/20220902-1-EN voices-be-heard-financial-institutions ———. 2023a. “Code of Banking Practice.” https://www. ———. 2023c. “Social Media Usage by Digital Finance hkma.gov.hk/eng/regulatory-resources/regulatory-guides/ Consumers: An Analysis of Consumer Complaints in Kenya and by-subject-current/code-of-banking-practice Nigeria.” Research Publication, Washington, D.C.: Innovations for Poverty Action. https://poverty-action.org/publication/social- ———. 2023b. “Consumer Protection in respect of Digital media-usage-digital-finance-consumers-analysis-consumer- Marketing Activities.” Circular, December 20. https://brdr.hkma. complaints-kenya-nigeria gov.hk/eng/doc-ldg/docId/20231220-1-EN Responsible Digital Credit: Frontier Solutions for Authorities and Providers 98 ITonDemand. 2018. “The Cost of Downtime: How Business Kenya Institute for Public Policy Research and Analysis (KIPPRA). Continuity Affects the Bottom Line.” https://itondemand. 2022. “Regulation of Digital Credit Providers in Kenya: Policy com/2018/05/29/costs-of-downtime/ Issues and Options.” Nairobi: KIPPRA. https://kippra.or.ke/ regulation-of-digital-credit-providers-in-kenya-policy-issues- Izaguirre, Juan Carlos. 2020. “Making Consumer Protection and-options/ Regulation More Customer-Centric.” Working Paper, Washington, D.C.: CGAP. https://www.cgap.org/research/ Kiva. 2022. “What’s Up with Microfinance Interest Rates?” publication/making-consumer-protection-regulation-more- https://www.kiva.org/blog/whats-up-with-microfinance- customer-centric interest-rates Izaguirre, Juan Carlos, Denise Dias, Eric Duflos, Laura Brix Klarna. 2023. “Rethinking Credit Reporting: A Call for Newbury, Olga Tomilova, and Myra Valenzuela. 2022a. “Market Transparency and Fairness.” https://www.klarna.com/ Monitoring for Financial Consumer Protection.” CGAP Toolkit, international/press/rethinking-credit-reporting-a-call-for- Washington, D.C.: CGAP. https://www.cgap.org/topics/ transparency-and-fairness collections/market-monitoring KPMG. 2015. “National Credit Factsheet.” South Africa, May. Izaguirre, Juan Carlos and Rafe Mazer. 2018. “How Regulators https://assets.kpmg.com/content/dam/kpmg/pdf/2016/05/ Can Foster More Responsible Digital Credit.” Blog, Washington, National-Credit-Factsheet-May%202015.pdf D.C.: CGAP. https://www.cgap.org/blog/how-regulators-can- foster-more-responsible-digital-credit Kruijff, David, Swati Sawhney, and Richard Leslie Wright. 2024. “Empowering Small Giants: Inclusive Embedded Finance for Izaguirre, Juan Carlos, Rafe Mazer, and Louis Graham. 2018. Micro-Retailers.” Focus Note, Washington, D.C.: CGAP. https:// “Digital Credit Market Monitoring in Tanzania.” Reading Deck, www.cgap.org/research/publication/empowering-small- Washington, D.C.: CGAP. https://www.cgap.org/research/ giants-inclusive-embedded-finance-for-micro-retailers reading-deck/digital-credit-market-monitoring-in-tanzania Lark Finserv. 2023. “eKYC and Its Importance in Digital Lending.” Izaguirre, Juan Carlos, Rafe Mazer, and Michelle Kaffenberger. https://www.larkfinserv.com/lark-resources/ekyc-and-its- 2018. “It’s Time to Slow Digital Credit’s Growth in East Africa.” importance-in-digital-lending Blog, Washington, D.C.: CGAP. https://www.cgap.org/blog/ its-time-to-slow-digital-credits-growth-in-east-africa Lea, Stephen. 2020. “Debt and Overindebtedness: Psychological Evidence and its Policy Implications.” Social Kaffenberger, Michelle. 2018. “Digital Credit in Tanzania: Issues and Policy Review 15: 146–179. https://doi.org/10.1111/ Customer Experiences and Emerging Risks.” Reading Deck, sipr.12074 Washington, D.C.: CGAP. https://www.cgap.org/research/ reading-deck/digital-credit-in-tanzania-customer- Likumi. 1999. “Patērētāju tiesību aizsardzības likums.” https:// experiences-and-emerging-risks likumi.lv/ta/en/en/id/23309-consumer-rights-protection-law Kaffenberger, Michelle, Edoardo Totolo, and Matthew Livshits, Igor. 2020. “Regulating Consumer Credit and Soursourian. 2018. “A Digital Credit Revolution: Insights from Protecting (Behavioral) Borrowers.” Economic Insights, Federal Borrowers in Kenya and Tanzania.” Working Paper, Washington, Reserve Bank of Philadelphia, 5 (1): 14–19. https://ideas.repec. D.C.: CGAP. https://www.cgap.org/research/publication/digital- org/a/fip/fedpei/87602.html credit-revolution-insights-borrowers-in-kenya-and-tanzania Maringa, Felix, and Bouba Jalloh. 2023. “Millions of Kenyans Kelly, Sonja, and Mehrdad Mirpourian. 2021. “Algorithmic Bias, indebted to mobile app lenders.” DW, April 17. https://www. Financial Inclusion, and Gender: A primer on opening up new dw.com/en/millions-of-kenyans-indebted-to-mobile-app- credit to women in emerging economies.” Women’s World lenders/a-65349197 Banking. https://www.womensworldbanking.org/insights/ Mazer, Rafe. 2018. “Does Transparency Matter?” Slide Deck, report-algorithmic-bias-financial-inclusion-and-gender- Washington, D.C.: CGAP. https://www.cgap.org/sites/default/ primer-opening-up-new-credit-to-women-in-emerging- files/publications/slidedeck/2018_03-Slidedeck-Does_ economies/ Transparency_Matter.pdf Responsible Digital Credit: Frontier Solutions for Authorities and Providers 99 Mazer, Rafe, and Alexandra Fiorillo. 2015. “Digital Credit: Ministry of Justice and Public Security. N.d. “Senacon.” https:// Consumer Protection for M-Shwari and M-Pawa Users.” Blog, www.gov.br/mj/pt-br/assuntos/seus-direitos/consumidor/ Washington, D.C.: CGAP. https://www.cgap.org/blog/digital- credit-consumer-protection-for-m-shwari-and-m-pawa-users Ministry of Law and Justice. 2016. “THE INSOLVENCY AND BANKRUPTCY CODE, 2016.” New Delhi, May 28. https://www. Mazer, Rafe, and Seth Garz. 2024. “Fast Growth and Slow mca.gov.in/Ministry/pdf/TheInsolvencyandBankruptcyofIndia. Policy: A Decade of Digital Credit in Kenya.” Oxford Review of pdf   Economic Policy 40 (1): 82–103. https://academic.oup.com/ oxrep/article-abstract/40/1/82/7630846 ———. 2023. “India Digital Personal Data Protection Act. No. 22.” New Delhi, August 11. https://www.meity.gov.in/static/ Mazer, Rafe, and Katharine McKee. 2017. “Consumer Protection uploads/2024/06/2bf1f0e9f04e6fb4f8fef35e82c42aa5.pdf in Digital Credit.” Focus Note, Washington, D.C.: CGAP. https:// www.cgap.org/research/publication/consumer-protection-in- MIT Sloan. 2023. “Digital Collateral.” https://mitsloan.mit.edu/ digital-credit sites/default/files/inline-files/Digital%20Collateral_QJE_ revision_2023-12-8.pdf ———. 2018. “Digital Consumer Credit: Four Ways Providers can Improve Customer Experience.” Slide Deck, Washington, Mitha, Ajaze, Faith Biegon, and Peter Zetterli. 2022. “Banking in D.C.: CGAP. https://www.cgap.org/research/reading-deck/ Layers.” Working Paper, Washington, D.C.: CGAP. https://www. digital-consumer-credit-four-ways-providers-can-improve- cgap.org/research/publication/banking-in-layers-five-cases- customer-experience to-illustrate-how-market-structure-for-financial Mazer, Rafe, Katharine McKee, and Alexandra Fiorillo. 2014. Monetary Authority of Singapore. 2017. “Credit Limit “Applying Behavioral Insights in Consumer Protection Policy.” Management Measure.” https://www.mas.gov.sg/regulation/ Focus Note, Washington, D.C.: CGAP. https://www.cgap. explainers/credit-limit-management-measure org/research/publication/applying-behavioral-insights-in- ———. 2021. “Granting Non-Card Unsecured Credit Facilities.” consumer-protection-policy https://www.mas.gov.sg/regulation/explainers/granting-non- Mazer, Rafe, Nikhil Ravichandar, and Julian Dyer. 2017. card-unsecured-credit-facilities “Can Digital Savings Reduce Risks in Digital Credit?” Blog, ———. 2024. “Guidelines on Shared Responsibility Washington, D.C.: CGAP. https://www.cgap.org/blog/ Framework.” https://www.mas.gov.sg/regulation/guidelines/ can-digital-savings-reduce-risks-in-digital-credit guidelines-on-shared-responsibility-framework Mazer, Rafe, James Vancel, and Ariana Keyman. 2016. “Finding Mustafa, Zeituna, Mercy Wachira, Vera Bersudskaya, William ‘Win-Win’ in Digitally-Delivered Consumer Credit.” Blog, Nanjero, and Graham A. N. Wright. 2017. “Where Credit Is Due: Washington, D.C.: CGAP. https://www.cgap.org/blog/finding- Customer Experience of Digital Credit In Kenya.” Microsave. win-win-in-digitally-delivered-consumer-credit https://www.microsave.net/wp-content/uploads/2018/10/ Medine, David, and Gayatri Murthy. 2020. “Making Data Work Where_Credit_Is_Due_Customer_Experience_of_Digital_ for the Poor: New Approaches to Data Protection and Privacy.” Credit_In_Kenya.pdf Washington, D.C.: CGAP. https://www.cgap.org/research/ National Anti-Financial Crime Center. N.d. “National Scam publication/making-data-work-for-poor Response Center (NSRC).” Malaysia. https://nfcc.jpm.gov.my/ Microsoft. 2024. “Encryption in Transit.” https://learn.microsoft. index.php/en/about-nsrc com/en-us/compliance/assurance/assurance-encryption-in- National Archives. 2013. “PART 1075—CONSUMER FINANCIAL transit CIVIL PENALTY FUND RULE.” Consumer Financial Protection Microsoft Press Store. 2021. “Authentication and Identity Bureau, May 7. https://www.ecfr.gov/current/title-12/chapter-X/ Protection in the Cloud.” https://www.microsoftpressstore. part-1075 com/articles/article.aspx?p=3172424 National Bureau of Economic Research (NBER). 2019. “Digital Microsoft Research and EY. 2020. “Fairlearn: Assessing Credit and Household Indebtedness.” Working Paper 25604, and Mitigating Unfairness in Credit Models.” https://www. Cambridge, MA: NBER. https://www.nber.org/papers/w25604 microsoft.com/en-us/research/wp-content/uploads/2020/09/ Fairlearn-EY_WhitePaper-2020-09-22.pdf Responsible Digital Credit: Frontier Solutions for Authorities and Providers 100 National Justice Council of Brazil. 2022. “Access to Justice ProjectPro. 2024. “Compare Microsoft Azure Active Directory for Consumers.” https://www.cnj.jus.br/wp-content/ vs. Google Cloud Firebase Authentication.” https://www. uploads/2022/09/access-justice-consumidor-gov-br.pdf. projectpro.io/compare/microsoft-azure-active-directory-ad- vs-google-cloud-firebase-authentication National Privacy Commission. 2020. “Guidelines On The Processing Of Personal Data For Loan-related Transactions.” Republic of Kenya. 2024. “The Business Laws (Amendment) NPC Circular No. 20-01, the Philippines, September 14. Act, 2024.” Kenya Gazette Supplement, Nairobi, December 13. https://privacy.gov.ph/wp-content/uploads/2020/10/ https://new.kenyalaw.org/akn/ke/act/2024/20/eng@2024-12- NPC-Circular-No.-20-01.pdf   13 Navarro, Antoine, Eric Duflos, and Juan Carlos Izaguirre. 2024. Republic of South Africa. 2006. “National Credit Act 34 of “The Stories Behind Late Repayment on Digital Credit in Cote 2005.” Government Gazette, Cape Town, March 15. https:// D’Ivoire.” Blog, Washington, D.C.: CGAP. https://www.cgap.org/ www.gov.za/documents/national-credit-act blog/stories-behind-late-repayment-on-digital-credit-in-cote- divoire ———. 2015. “National Credit Act: Regulations.” Government Gazette, March 13. https://www.gov.za/documents/notices/ Nkhonjera, Maria. 2017. “Competition Authority of Kenya (CAK) national-credit-act-regulations-13-mar-2015 rules on USSD Pricing.” Blog. https://www.competition.org. za/ccred-blog-competition-review/2017/4/12/competition- Reserve Bank Innovation Hub (RBIH). 2024. “Unified Lending authority-of-kenya-cak-rules-on-ussd-pricing Interface: Project Overview.” Bengaluru. https://rbihub.in/ unified-lending-interface/ OECD. 2019a. “Recommendation of the Council on Consumer Protection in the field of Consumer Credit.” https:// Reserve Bank of India (RBI). 2021. “The Reserve Bank – legalinstruments.oecd.org/en/instruments/oecd-legal-0453 Integrated Ombudsman Scheme, 2021.”  https://rbidocs.rbi.org. in/rdocs/content/pdfs/RBIOS2021_amendments05082022. ———. 2019b. “Short-Term Consumer Credit: Provision, pdf Regulatory Coverage, and Policy Responses.” https:// web-archive.oecd.org/2022-06-21/507614-Short-term- ———. 2022a. “Guidelines on Digital Lending.” consumer-credit-report.pdf September 2. https://rbi.org.in/scripts/NotificationUser. aspx?Id=12382&Mode=0 ———. 2022. “Updated G20/OECD High-Level Principles on Financial Consumer Protection.” https://www. ———. 2022b. “Master Direction – Reserve Bank of India oecd.org/finance/high-level-principles-on-financial- (Securitisation of Standard Assets) Directions, 2021.” December consumerprotection.htm 5. https://www.rbi.org.in/Scripts/BS_ViewMasDirections. aspx?id=12165 ———. 2024. “Consumer Finance Risk Monitor.” Paris: OECD Publishing. https://doi.org/10.1787/047b2ea6-en ———. 2023. “RBI Consumer Grievance Redressal Scheme.” https://www.rbi.org.in/commonperson/English/Scripts/FAQs. OpenSense Labs. 2023. “AI Fairness: Understanding Bias and aspx?Id=3407#:~:text=The%20Scheme%20simplifies%20 Ensuring Ethical AI Systems.” https://opensenselabs.com/blog/ the%20grievance,at%20one%20centralised%20reference%20 ai-fairness point Payment Systems Regulator. 2023. “Policy statement: Fighting Robinson, Jonathan, David Sungho Park, and Joshua E. authorized push payment scams: final decision.” UK. https:// Blumenstock. 2022. “The Impact of Digital Credit in Developing www.psr.org.uk/media/kwlgyzti/ps23-4-app-scams-policy- Economies: A Review of Recent Evidence.” Berkeley, CA: Center statement-dec-2023.pdf for Effective Global Action. https://dshpark.com/files/DCO_ synthesis.pdf#page=24&zoom=100,96,402 PivIT Global. 2021. “How Much Does IT Downtime Really Cost?” https://info.pivitglobal.com/blog/cost-of-downtime RSA. 2023. “Understanding MFA Methods: Security Keys, Tokens and Beyond.” https://www.rsa.com/resources/blog/ Plaid. 2023. “Data Transparency Messaging Migration Guide.” multi-factor-authentication/understanding-mfa-methods- https://plaid.com/docs/link/data-transparency-messaging- security-keys-tokens-and-beyond/ migration-guide/ Safaricom. 2024a. “Data Privacy Statements.” https://www. safaricom.co.ke/data-privacy-statements Responsible Digital Credit: Frontier Solutions for Authorities and Providers 101 ———. 2024b. “M-Shwari FAQs.” https://www.safaricom. Tala. N.d. “Home.” Accessed April 28, 2025. https://tala.co/ co.ke/media-center-landing/frequently-asked-questions/m- shwari-faqs-safaricom Tala. 2020. “Leveraging AWS Lambda to Drive Product Growth.” https://tala.co/blog/2020/01/14/leveraging-aws-lambda-to- ———. 2024c. “M-Shwari Terms and Conditions.” https://www. drive-product-growth/ safaricom.co.ke/media-center-landing/terms-and-conditions/ m-shwari-terms-conditions#:~:text=14.,out%20in%20the%20 ———. 2021. “Helping Late Borrowers Repay with Uplift Privacy%20Policy. Modeling.” Blog. https://tala.co/blog/2021/01/14/helping-late- borrowers-repay-with-uplift-modeling/ ———. 2024d. “Position Statement on Data Privacy.” https:// www.safaricom.co.ke/media-center-landing/press-releases/ ———. 2024. “Data Ethics Policy.” https://tala.co/data-ethics/ position-statement-data-privacy Tala Kenya. 2024a. “App Privacy Notice.” https://tala.co.ke/ ———. 2024e. “Zuri – Interactive Customer Care Chatbot.” app-privacy-notice/ https://www.safaricom.co.ke/personal/value-added-services/ ———. 2024b. Tala Kenya Homepage. https://tala.co.ke/ information-services/zuri Tekedia. 2021. “M-Pesa Outages: The Implication on the Kenyan Semafor. 2023. “Google Is Africa’s Fintech Loan Regulator.” Economy and the Need for More Mobile Money Platforms to https://www.semafor.com/article/02/06/2023/google-is- Emerge.” https://www.tekedia.com/m-pesa-outages-the- africas-fintech-loan-regulator implication-on-the-kenyan-economy-and-the-need-for-more- Singapore FinTech Association (SFA). 2022. “Buy Now, Pay Later mobile-money-platforms-to-emerge/ Code of Conduct.” Singapore: SFA. https://singaporefintech. Thailand Consumers Council. 2025. “Consumer Council org/wp-content/uploads/2023/05/BNPL-CoC-Final-170822- reveals 6 solutions to solve OPPO – Realme problems.” https:// Code.pdf www-tcc-or-th.translate.goog/6-consumerpolicies-oppo- Spriggs, Rebecca, and Kennedy Kipkemboi. 2024. “Financial realme/?_x_tr_hl=en&_x_tr_pto=wapp&_x_tr_sl=auto&_x_tr_ Health: Addressing Consumer Overindebtedness in the tl=en Digital Age.” Report, GSMA. https://www.gsma.com/solutions- Thales Group. 2024. “Cloud Key Management.” https://cpl. and-impact/connectivity-for-good/mobile-for-development/ thalesgroup.com/encryption/key-management/cloud-key- wp-content/uploads/2024/12/Financial-Health-Report-2024. management pdf Tomilova, Olga, and Myra Valenzuela. 2018. “Financial Inclusion Standard Media. 2020. “Digital Lenders Cut Credit to + Stability, Integrity, and Protection (I-SIP): Policy Making for Kenyans After CBK Directive.” September 12. https://www. an Inclusive Financial System.” Technical Guide, Washington, standardmedia.co.ke/business/news/article/2001430705/ D.C.: CGAP. https://www.cgap.org/research/publication/i-sip- digital-lenders-cut-credit-to-kenyans-after-cbk-directive toolkit-policy-making-for-inclusive-financial-system Stewart, Amy, Kathleen Yaworsky, and Paul Lamont. 2018. Traders Union. 2024. “Kubo Financiero Review.” https:// “Demystifying Digital Lending: How Digital Transformation tradersunion.com/reviews/kubofinanciero-com/ Can Help Financial Service Providers Reach New Customers, Drive Engagement, and Promote Financial Inclusion.” Accion, TrustDecision. 2024. “Detecting Application Fraud on an Cambridge, MA. https://www.accion.org/wp-content/ Indonesian Cash Loan Platform.” https://trustdecision.com/ uploads/2018/09/1123_Digital-Lending_R10_Print_Ready-2.pdf resources/blog/case-study-detecting-application-fraud-on- an-indonesian-cash-loan-platform Suri, Tavneet, Prashant Bharadwaj, and William Jack. 2021. “Fintech and household resilience to shocks: Evidence from Twilio. 2024a. “SMS Messaging Channel.” https://www.twilio. digital loans in Kenya.” Journal of Development Economics 153: com/en-us/messaging/channels/sms 102697. https://www.sciencedirect.com/science/article/pii/ S0304387821000742 ———. 2024b. “Fraud Prevention Use Cases.” https://www. twilio.com/en-us/use-cases/fraud-prevention Suriani, Fernanda. 2022. “Access to justice and Consumidor.gov. br case.” https://www.cnj.jus.br/wp-content/uploads/2022/09/ Uganda Microfinance Regulatory Authority. 2024. “Digital access-justice-consumidor-gov-br.pdf Lending Guidelines.” https://umra.go.ug/digital-lending- guidelines/ Responsible Digital Credit: Frontier Solutions for Authorities and Providers 102 UNCTAD. 2024. “Consumer Dispute Resolution in the Wawad, Joud. 2023. “AWS KMS Deep Dive.” https://medium. World.” https://unctad.org/system/files/official-document/ com/@joudwawad/aws-kms-deep-dive-b482808d5f63 ditccplp2023d2_en.pdf World Bank. Forthcoming. “Guidance Note: Policy Options for University of Johannesburg. 2021. “The Impact of the National Financial Consumer Over-Indebtedness.” World Bank Group, Credit Act on Consumer Lending.” UJ Content. https:// Washington, D.C. ujcontent.uj.ac.za/esploro/outputs/graduate/The-impact-of- the-National-Credit/9910651307691 World Bank Group. 2017. “Good Practices for Financial Consumer Protection. 2017 Edition.” World Bank Group, US Code of Federal Regulations. 2013. “Title 12 Chapter X Part Washington, D.C. https://openknowledge.worldbank.org/ 1075, CONSUMER FINANCIAL CIVIL PENALTY FUND RULE.” entities/publication/ff9c9482-5ffe-5c98-b6f3-5b97eeab5ed1 https://www.ecfr.gov/current/title-12/chapter-X/part-1075 ———. 2023. “Good Practices for Credit Reporting.” Doing US House of Representatives. 1978. “TITLE 11—BANKRUPTCY.” Business 2023. https://subnational.doingbusiness.org/en/data/ Washington, D.C., November 6. https://uscode.house.gov/view. exploretopics/getting-credit/good-practices#GP-CI4 xhtml?edition=prelim&path=%2Fprelim%40title11 Yubico. 2024. “Finance Industry Solutions.” https://www.yubico. ValidSoft. 2024. “ValidSoft Voice Authentication Solutions.” com/industries/finance https://www.validsoft.com/ Zendesk. 2023. “Understanding Chatbot Options in Your Venkatesan, Jayshree, Rafe Mazer, and Collin Rice. 2024. Zendesk Account.” https://support.zendesk.com/hc/en-us/ “Positive Friction for Responsible Digital Lending: A Call to articles/4408833172250-Understanding-chatbot-options-in- Action.” Center for Financial Inclusion, USAID Digital, and DAI your-Zendesk-account Global, February. https://www.centerforfinancialinclusion.org/ positive-friction-for-responsible-digital-lending-a-call-to- ———. 2024. “About Automated Resolutions for action/ AI Agents.” https://support.zendesk.com/hc/en-us/ articles/8357756668186-About-automated-resolutions-for- VentureBeat. 2021. “How Voice Biometrics Is Saving Financial AI-agents Services Companies Millions and Eliminating Fraud.” https:// venturebeat.com/business/how-voice-biometrics-is-saving- financial-services-companies-millions-and-eliminating-fraud/ Responsible Digital Credit: Frontier Solutions for Authorities and Providers 103 CGAP Members CGAP members and strategic partners as of June 2025 CGAP Members (continued) CGAP Strategic Partners CGAP members and strategic partners as of June 2025 Transforming Lives with Financial Inclusion cgap.org