## pffiea ## Source details **Canonical URL:** [pffiea](https://www.imf.org/-/media/files/publications/dp/2020/english/pffiea.pdf) ## Other formats - [Markdown version](/-/media/files/publications/dp/2020/english/pffiea.pdf.md) - [Structured JSON version](/-/media/files/publications/dp/2020/english/pffiea.pdf.json) --- ### Executive summary and overview - Technology-driven change in financial services has increased access globally and accelerated during the COVID-19 pandemic. - New index of digital financial inclusion introduced; analysis covers a sample of 52 EMDEs and combines quantitative analysis (pre-COVID data) with interviews of stakeholders (representatives of more than 70 fintech companies, central banks, regulatory bodies, and banks). - Focus: payments and credit (savings, insurance, and wealth management remain nascent and data-poor). ### Key inclusion and reach findings - Globally 1.7 billion people have no access to a bank account. - Small- and medium-sized enterprises (SMEs) constitute 95 percent of companies worldwide and provide employment to more than 60 percent of workers. - In all the 52 countries covered, digital financial inclusion increased between 2014 and 2017, even where traditional financial inclusion stalled or declined. - Digital financial inclusion is evolving from “spend” to “lend,” filling gaps where traditional delivery is less present. - Digital financial services are faster, more efficient, and typically cheaper than traditional services and increasingly reach lower-income households and SMEs. - Mobile money and digital payments indicators and exact figures: - The number of active mobile money accounts almost tripled and use of mobile phone for domestic remittances roughly doubled between 2013 and 2017 in lower-middle and low-income countries. - In low-income countries, about half of the population received or sent remittances using mobile phones in 2017. - Transactions in Cambodia, Ghana, and Zimbabwe reached more than 75 percent of GDP in 2018. - The share of adults making or receiving digital payments increased by 11 percentage points between 2014 and 2017, to reach 52 percent. - In 2017, more than 2.9 million mobile money agents operating in 90 countries facilitated cash-in/cash-out transactions, peer-to-peer transfers, and bill payments. - Smartphones increasingly affordable at US$20 to US$30. ### Marketplace lending and fintech credit - Marketplace lending doubled from 2015 to 2017. - By 2017, marketplace lending reached US$400 billion. - Marketplace lending volume remains very small, at less than 0.5 percent of GDP for most countries. - In 2017, fintech credit was dominated by China, followed by the United States and the United Kingdom—together they made up 98 percent of the fintech credit market. - Marketplace lending is mostly comprised of very small loans likely to small borrowers. ### Macroeconomic implications and empirical results - Empirical analysis (cross-country OLS for 52 countries) finds a significant positive association between digital financial inclusion and GDP growth; traditional financial inclusion does not show the same consistently positive association. - Key OLS coefficient patterns for growth regressions (Digital F.I Index and Traditional F.I Index across pooled specifications): - Digital F.I Index coefficients (as reported across pooled columns): 0.272 *, 0.364 ***, 0.376 ***, 0.311 **, 0.314 **, 0.371 ***, 0.175 **, 0.150 **, 0.148 **, 0.129 *, 0.130 *, 0.188 **. - Traditional F.I Index coefficients (as reported across pooled columns): 0.157, 0.241 **, 0.176, 0.129, 0.004, 0.144, 0.0606, 0.109, 0.112, 0.024, 0.023, 0.105. - Other consistent regression results: - Log (GDP per capita) is positive and significant across specifications (examples: 0.0879 ***, 0.0940 ***, 0.0880 ***, …). - Government consumption (% of GDP) often positive and significant (example coefficients include 0.016 *** in several specifications). - Log (private credit) (% of GDP) is positive and significant in several specifications (examples: 0.0401 **, 0.0504 **). ### Determinants of digital financial inclusion (access and usage) - Digital and traditional components combined via a three-stage PCA to build indices for 52 EMDEs (2014–17). - Enablers identified: - Customer identification and digital ID (example: Aadhaar in India used by more than 1 billion people). - Digital infrastructure: internet and mobile phone access strongly associated with higher usage of digital payments and credit. - Financial literacy and familiarity with traditional services: higher traditional usage is strongly associated with higher fintech usage. - Supportive regulatory and legal environment, and availability of credit information and stronger legal rights support digital lending. - Empirical determinants and coefficient highlights (pooled regressions and fractional logit): - Internet users (% of population) positively associated with marketplace lending (example coefficients: 3.541 ***, 3.539 ***, 3.786 ***, … in Annex regression table). - Mobile money agent density (lagged) positively associated with fintech usage; reported coefficient values in fractional logit results include many significant estimates in the 0.00166–0.00302 range (e.g., 0.00302 ***). - Overhead cost to assets (bank inefficiency) positively associated with fintech access (examples: 28.82 **, 43.95 ***, 51.91 ***). - Bank concentration negatively associated with fintech access (examples: −3.817 **, −3.273 **). - Log (NPL) often positively associated with fintech access (example: 94.18 ***). - PCA loadings and variance explained (selected values preserved): - Access (Traditional) PC1: ATM per 100,000 population 0.7071; bank per 100,000 population 0.7071. - Usage (Fintech-driven) PC1: Mobile account (%) 0.5130; Mobile for wages (%) 0.5356; Mobile for utility (%) 0.5580. - Cumulative variance examples: Access (Traditional) PC1 0.7982, Usage (Traditional) PC1 0.7759, Traditional Financial Inclusion Index PC1 0.8448, Overall Financial Inclusion Index PC1 0.6083. ### COVID-19 role, short- and long-term implications - Digital financial services have enabled contactless transactions and efficient deployment of government support (G2P), with mobile money used for transfers where banking access is limited (examples cited: Namibia, Peru, Uganda, Zambia). - Short-term empirical anecdotes: - Mobile money transactions in Rwanda increased by 450 percent between January and April 2020. - Number of users sending money virtually in Rwanda: 0.6 million in the week before lockdown, 1.2 million in the week after lockdown, and 1.8 million in the final week of April. - Download of finance mobile applications estimated to have increased by 24 percent on average across 74 countries since the start of the lockdown. - Short-term risks: - Tighter funding conditions and collapse in consumption have strained fintechs: fintech funding activity stalled in Q1 2020 (Asia saw a 69 percent drop in funding and a 23 percent drop in deals quarter-over-quarter). - Smaller fintechs with thinner buffers are at risk; consolidation could increase market concentration and reduce access for small customers. - Rapid shift to digital services risks creating new exclusion for those without mobile phones, computers, internet, or adequate digital/financial literacy. - Long-term potential: - COVID-19 could accelerate digital financial inclusion (historical precedent: SARS in 2003 and China’s digital payments uptake), but outcomes depend on digital infrastructure scaling and policy support. ### Risks to inclusion and financial stability - Direct risks: - Technology-related risks: data leakage, cyberattacks. - ML/TF and privacy concerns; algorithmic biases may entrench exclusion. - Predatory or inappropriate lending by underregulated providers could undermine trust. - Minimum “tech capital” and human capital requirements can exclude women, the poor, and the elderly. - Indirect and systemic risks: - Regulatory arbitrage, disruption of traditional business models, and interconnectedness between banks and lightly supervised fintech firms could pose spillover risks. - Automation of credit decisions could increase procyclicality and lead to faster contraction in downturns. - Microfinance institutions and small banks may face competitive pressure and operational strain, risking increased exclusion if they retrench before digital alternatives scale sufficiently. - User protection and safeguarding customer funds: - Risks from operator bankruptcy or failure of banks holding pooled e-float balances; mitigation options listed include segregation of customer funds, investing funds in safe liquid assets diversified across safe banks, or requiring deposits at the central bank (examples: China, Peru, the Philippines, Thailand require e-float deposited at the central bank—referred to as synthetic CBDC, sCBDC). ### Policy implications and recommendations - Regulatory and supervisory actions: - Adapt regulatory approaches to balance enabling innovation and addressing financial integrity, consumer protection, and financial stability. - Supervisory and regulatory frameworks should keep pace with accelerated growth of digital financial services. - Tools include innovation hubs, regulatory sandboxes, and adoption of RegTech and SupTech. - Domestic policy priorities: - Protect data and privacy; prevent cyber risk; facilitate digital infrastructure (electricity, mobile, internet); strengthen legal and institutional frameworks (digital ID, customer identification); ensure standardization and interoperability. - Promote consumer protection, contingency planning, and financial and digital literacy. - Fiscal measures to reach vulnerable populations should be integrated with investments in digital infrastructure and literacy. - International cooperation: - International agreements needed on data privacy, cybersecurity, digital identification, cross-border digital currencies, anti-trust to preserve competition, and regulation of Big Techs. - Global cooperation is necessary to mitigate risks of global monopolies, regulatory arbitrage, cross-border activities, cybersecurity, and money laundering. - Political-economy and distributional considerations: - Fintech can help close gender gaps, but barriers remain (access to resources, cultural norms, literacy). - Policymakers should consider distributional impacts: disappearance of smaller fintechs with high reach to low-income households could increase income divide and carry political/social costs. ### Data, methodological limitations, and outreach - Data limitations: - Index covers only 52 EMDEs (2014–17 for digital component); excludes advanced economies and the COVID-19 period. - Data do not fully capture savings, insurance, wealth management, or cross-border payments services; raw data do not distinguish provider type (incumbent bank vs fintech). - Short time series and limited instruments constrain causal identification. - Methodology: - Three-stage PCA used to build Access and Usage sub-indices for traditional and fintech-driven components, then combined into comprehensive financial inclusion index; variables winsorized at 2nd and 98th percentiles and normalized between 0 and 1. - Outreach: - Interviews with more than 70 fintech companies, and with central banks and regulators (including the Bank of England, central banks of China, Indonesia, Kenya, Morocco, Rwanda, Sierra Leone, Uganda) and organizations listed in Annex 1. *Source: IMF staff (pffiea).* ### Executive Summary ������������������������������������������������������������������������������������������������������ ### Executive Summary ### Overview - Technology is changing the landscape of the financial sector, increasing access to financial services in profound ways; these changes have affected nearly all countries in the world and have accelerated during the COVID-19 pandemic. - The paper introduces a new index of digital financial inclusion and shows that digital finance is increasing financial inclusion and is associated with higher GDP growth. - Interviews were conducted with stakeholders including representatives of more than 70 fintech companies, central banks, regulatory bodies, and banks to supplement quantitative analysis. ### Key findings on digital financial inclusion and reach - Globally 1.7 billion people have no access to a bank account. - Small- and medium-sized enterprises (SMEs) constitute 95 percent of companies worldwide and provide employment to more than 60 percent of workers, yet they struggle to access finance. - In all the 52 countries covered in the analysis, digital financial inclusion increased between 2014 and 2017, even where traditional financial inclusion was stalling or declining. - Digital financial inclusion is evolving from “spend” to “lend,” and tends to fill gaps where traditional delivery of financial services is less present. - Digital financial services are faster, more efficient, and typically cheaper than traditional financial services, increasingly reaching lower-income households and SMEs. ### Macroeconomic implications - Empirical analysis shows a positive association between digital financial inclusion and GDP growth, using an approach that addresses potential reverse causality. - Digital financial inclusion could play an important role in mitigating the economic and social impact of the COVID-19 crisis and support a more inclusive recovery. ### Role during COVID-19 - During the COVID-19 health crisis, digital financial services have enabled contactless and cashless transactions and facilitated efficient and quick deployment of government support measures. - Where access to banking networks is limited, mobile money networks are being used to deliver government transfers (examples cited include Namibia, Peru, Uganda, Zambia). - Fintech has helped connect governments to informal workers outside formal benefits programs and supported tax authorities encouraging online platform use (examples cited include Kenya, Namibia, Nigeria). - Some fintech lenders responded to SME liquidity needs quickly (example cited: China), and many fintech companies offered flexibility in loan repayments (examples cited: India, Kenya, United Kingdom). - The fiscal response should go hand-in-hand with investment in digital infrastructure and promoting digital and financial literacy to ensure greater digital inclusion. ### Inclusion outcomes and distributional effects - Digital financial inclusion appears to be closing gender gaps, though barriers remain: access to resources (mobile phone, internet), cultural or social norms, and digital and financial literacy may be higher for women. - Lack of access to mobile phones, computers, or the internet could lead to new forms of exclusion as the shift to digital financial services accelerates. ### Industry structure, competition, and collaboration - Delivery of digital financial services is evolving with various models of interaction between incumbents and disruptors. - Fintech companies frequently are the source of innovation and often compete with banks, which are responding by investing heavily in fintech. - Widespread collaboration based on complementarities between fintech firms and financial institutions is also observed. - The COVID-19 crisis is a test of fintech resilience; tightening funding conditions and a sharp drop in transactions are already affecting fintech companies, especially smaller ones and those with thinner buffers. - Widespread consolidation and retrenchment by smaller companies could lead to greater concentration and reduce financial access for small customers. ### Risks to financial inclusion and financial stability - Risks to financial stability include regulatory arbitrage leading financial activities to migrate to less or lightly regulated sectors, disruption of traditional business models, and interconnectedness between traditional institutions and lightly supervised fintech firms. - Technology-related risks (data leakage, cyberattacks) and ML/TF risks are present for both banks and nonbank financial institutions. - Inappropriate lending practices by underregulated institutions could jeopardize trust. - Unequal access to digital infrastructure and potential biases amplified by new data sources and analytics could create or exacerbate financial exclusion. ### Policy implications and recommendations - Policymakers should adapt regulatory approaches to strike a balance between enabling financial innovation and addressing challenges and risks to financial integrity, consumer protection, and financial stability. - Supervisory and regulatory frameworks should keep pace with the accelerated growth of digital financial services to mitigate financial stability risks. - International agreements and cooperation are needed on data privacy, cybersecurity, digital identification, cross-border digital currencies, and regulation of Big Techs to preserve competition in the post-COVID era. - Fiscal measures aiming to reach vulnerable populations should be integrated with investments in digital infrastructure and efforts to promote digital and financial literacy. _Executive Summary — The Promise of Fintech: Financial Inclusion in the Post COVID-19 Era_ ### Introduction ### Introduction ### Overview - During the COVID-19 crisis, smooth access to government electronic systems integrated with digital financial services platforms (fintech firms, mobile money companies, and digital banking) is critical to provide wide-reaching policy support promptly and without contact to the public. - If government electronic systems are not easily accessible or well integrated, fiscal support announcements—even large ones—will fail to reach the most vulnerable and needy. - Fintech companies identified limited supply of skilled labor and digital infrastructure as major constraints. - Data biases or inaccurate and insufficient information could increase financial exclusion and feed distrust for new technology, especially among the most vulnerable. Lack of financial and digital literacy could exacerbate these risks. - Financial inclusion could be threatened by the possible demise of microfinance institutions, whose operations and clients might be affected more by the economic fallout and who might be struggling to operate digitally during the COVID-19 crisis. ### Fintech sector stresses during COVID-19 - Many fintech companies are young and have never experienced an economic downturn before; the COVID-19 crisis is the first major test of the fintech sector’s resilience. - Tighter funding conditions will affect fintech companies with thinner liquidity buffers: - Fintech funding activity stalled in the first quarter of 2020, resulting in the worst first quarter for fintech funding since 2017, as investors pulled back investments. - The funding draught could lead to smaller fintech firms being bought by larger firms or disappearing, potentially increasing market concentration in the fintech sector. - The economic crisis and collapse in consumption in heavily impacted sectors (hotels, restaurant, airlines, and retail) will cause a fall in fintech payments firms’ revenues. - Much fintech lending has targeted small borrowers, who are likely to be disproportionately affected and hence may see a sharp deterioration in loan quality. - Major disruptions to services provided by fintech companies could set back progress in digital financial inclusion and produce macroeconomic and financial spillovers. ### Focus and approach of the analysis - The paper uses a two-pronged approach given data limitations: - Quantitative analysis based on available pre-COVID-19 data. - Information from interviews around the world with policymakers, regulators, fintech companies, and banks. - The work focuses on two leading aspects of financial inclusion: access to domestic payments and credit. Savings, insurance, and wealth management are still nascent and data-poor. - Cross-border payments (notably remittances) have high fintech potential; global remittance cost remains at almost 7 percent, above the 3 percent target set by the SDGs. Support for remittances is noted but outside the scope of this note. - The paper introduces a new index to measure digital financial inclusion and compares it with an index of traditional financial inclusion for a sample of 52 emerging market and developing economies (EMDEs) prior to the COVID-19 crisis. - Global developments in marketplace lending (a subset of digital credit) are analyzed. - Determinants of digital financial inclusion and its impact on economic growth are assessed. - Interviews: representatives of more than 70 fintech companies, central banks, regulatory bodies, and banks were interviewed in two phases: - Phase 1: before COVID-19, providing insights on fintech potential for financial inclusion, competitive landscape, impediments, role of regulation, and risks. - Phase 2: follow-up interviews with a subsample to understand COVID-19 impacts on digital financial inclusion, business and clients, responses and collaborations, and future roles. ### Key findings - Digital finance is increasing financial inclusion, even where traditional financial inclusion is declining: - In all 52 countries covered, digital financial inclusion improved between 2014 and 2017, particularly in Africa and Asia, and even where traditional financial inclusion was stalling or declining. - In a sample of more than 100 countries, marketplace lending grew fast, albeit from a small base. - Digital financial inclusion tends to develop where traditional delivery of financial services is less present. - Interview evidence shows mixed effects on digital credit during COVID-19: in some countries fintech lenders participate in government schemes to support SME credit, whereas in others many fintech firms are scaling down new lending to preserve liquidity and manage credit risks. - Digital financial inclusion is associated with higher GDP growth: - Adoption of digital payments is significantly and positively associated with growth. - Fintech could mitigate the economic impact of COVID-19 and support recovery by helping ensure continued access to financial services, delivering government support effectively and securely, and supporting consumption, innovation, and productivity through digital economy developments. - However, potential disappearance of smaller fintech companies with higher reach to low-income households and small businesses could increase the income divide between rich and poor. - Fintech is contributing toward closing gender gaps in financial inclusion in most countries, but there is concern these gaps may rise in the post-COVID era: - Stakeholders cited higher barriers for women: cultural or social norms, financial literacy, safety, and disparity in access to resources. - Delivery models are evolving with interaction between incumbents and disruptors: - Fintech firms often compete with banks; established institutions are investing in fintech. - Widespread collaboration exists based on complementarities. - COVID-19 could increase both collaboration and competition as traditional institutions accelerate digitization; policy measures that route support through the banking sector could affect the competitive landscape. - Consolidation driven by tighter funding for smaller fintech firms could lead to greater sector concentration. - COVID-19 illustrates opportunities for governments to collaborate with private digital financial service providers to increase the reach of e-government services. - Safe development of digital financial inclusion depends on multiple factors: - Rapid inclusion without proper regulation and financial literacy can lead to financial instability. - Regulators emphasize cybersecurity risks, inappropriate lending practices by underregulated institutions, and the need for consumer protection, digital identification, and financial/digital literacy. - Fintech companies highlight skilled labor supply and availability of digital financial infrastructure as major constraints. - Digital finance can create new risks to financial inclusion: - Unequal access to digital infrastructure, constraints to financial and digital literacy, and potential biases amplified by new data sources and analytics. - Lightly regulated digital lending could threaten financial stability. - Indirect risks include disruption to microfinance institutions and loss of trust in digital technology. - These risks are amplified by the rapid shift toward digital financial services during COVID-19. ### Data limitations and coverage - Data constraints affecting analysis: - Data do not fully capture savings/wealth management and insurance instruments and many aspects related to credit; they exclude cross-border payments services. - Detailed digital payments data are available only for a relatively small sample of countries (52 out of about 190) and exclude the period from the onset of the COVID-19 health crisis as well as advanced economies. - Databases used to capture digital services do not identify the provider; such data could reflect services provided by incumbents, fintech companies, or other private sector providers. *Source: pffiea - Introduction* ### Introduction ### pffiea - Introduction ### Key limitations and scope of the analysis - Empirical analysis limitations: - Available time series is short and excludes several components of digital finance, and therefore likely underestimates the impact on growth. - Determining causality between growth and fintech-related financial inclusion remains a challenge; short time span of data and lack of good instruments constrain addressing endogeneity/reverse causality. - Interview sample may not be sufficiently representative of the population. - Scope of the paper: - Focuses on payments and credit among the many dimensions of financial services affected by fintech. - Organized around six broad questions: Where was digital financial inclusion emerging prior to the COVID-19 crisis? Is fintech increasing financial inclusion? What are the macroeconomic effects of digital financial inclusion? Are fintech companies disrupting traditional providers and how could these relationships evolve during the downturn and recovery? What are the impediments to the development of digital financial inclusion? What are the risks of fintech to financial inclusion? - Three fundamental technological changes driving fintech: - Massive data generation. - Advances in computer algorithms. - Increases in processing power, facilitated by high-speed broadband internet, cloud computing, and artificial intelligence; enabling big data analytics, blockchain technology, and biometric identification. ### Setting and recent international developments - Historical and institutional milestones: - Alliance for Financial Inclusion (AFI) started in 2008; Maya Declaration in 2011. - G20 Seoul Summit 2010 endorsed a Financial Inclusion Action Plan (FIAP) and created the Global Partnership for Financial Inclusion (GPFI). - United Nations adopted SDGs for 2030 in 2015, with financial inclusion featuring prominently. - AFI and GPFI created a Fintech for Financial Inclusion workstream in 2016. - IMF and World Bank launched the Bali Fintech Agenda in 2018. - COVID-19 pandemic: - Has amplified the benefits of expanding digital financial services and put focus on digital financial inclusion as a policy tool during the crisis. ### Literature review and empirical landscape - Data and composite measures: - Launch of IMF’s Financial Access Survey (FAS) and World Bank’s Global Findex enabled multidimensional composite measures of financial inclusion. - Two parametric approaches used for constructing composite indices: principal component analysis (PCA) and common factor analysis. - Empirical and regional studies: - Country- and region-specific work on mobile money (e.g., Kenya) and regional fintech developments (sub-Saharan Africa; Latin America and the Caribbean; Pacific Islands; Middle East and Central Asia). - Determinants of mobile money adoption often linked to GDP growth, levels of per capita income, regulatory environment, and rule of law. - Role of a lead firm (e.g., Ant Financial Services Group in China) is recognized. - Digital lending studies: - Determinants include regulation, financial development, digital infrastructure, and market structure. - Evidence suggests fintech lenders can process mortgage applications faster, reach underserved customers, and offer lower-cost credit than traditional lenders. ### Risks and emerging regulatory/policy concerns - Early-stage analysis on stability and inclusion risks: - Concerns include privacy, ML/TF risks, potential macroeconomic impact of digital currencies. - Financial stability issues from increased interconnectedness between fintech companies and banks and growing digital credit origination. - Risk of fintech-induced exclusion due to lack of access to digital infrastructure, differences in financial and digital literacy, or algorithmic biases—risks heightened by abrupt shift to digital services amid COVID-19 (including for G2P payments). ### Where fintech for financial inclusion is emerging - Market dynamics: - Fintech often fills gaps left by traditional institutions that rely on brick-and-mortar networks and legacy technology. - Fintech firms, including some from Silicon Valley, the United Kingdom, and China, are expanding into emerging markets such as India, Kenya, Mexico, Nigeria, and Tanzania. - Fintech firms often target niche markets and lower-income groups using the latest technology and data analytics. - Payments businesses have low (or negative) profitability; fintech firms are moving from “spend” into “lend” to improve viability. ### Spend: digital payments—evidence and trends - Role and advantages: - Mobile money and digital payments reduce need for face-to-face interaction and cash handling; beneficial for social distancing during COVID-19. - Mobile money can operate on basic feature phones; smartphones increasingly affordable at US$20 to US$30. - Regional leaders: - Largest increases in digital payments seen in Africa and Asia, with East Africa, China, and India leading. - Mobile money has cut remittance costs by 50 percent in parts of Africa (GSMA 2016, using the World Bank’s Remittance Prices Worldwide database). - Empirical indicators and exact figures: - The number of active mobile money accounts almost tripled and the use of mobile phone for domestic remittances roughly doubled between 2013 and 2017 in lower-middle and low-income countries. - In low-income countries, about half of the population received or sent remittances using mobile phones in 2017. - Transactions in Cambodia, Ghana, and Zimbabwe reached more than 75 percent of GDP in 2018. - The share of adults making or receiving digital payments increased by 11 percentage points between 2014 and 2017, to reach 52 percent (Demirgüç-Kunt and others 2018). - In 2017, more than 2.9 million mobile money agents operating in 90 countries facilitated cash-in/cash-out transactions, peer-to-peer transfers, and bill payments (GSMA 2018). - Policy responses accelerating adoption: - COVID-19 related measures include lowering fees and increasing limits on mobile money transactions (examples cited: Kenya, Uganda, Zambia), easing KYC for small transactions (example: Ghana), and relaxing interoperability rules (example: Democratic Republic of Congo). - Governments collaborating with fintech providers to expand G2P channels (example: Peru expanding providers to include private banks and mobile money providers). ### Lend: from payment data to microcredit - Mechanism and market developments: - Digital payments generate alternative data that fintech lenders use with “loan engines” (innovative algorithms) to identify credit-worthy clients and provide mostly unsecured lending. - Ant Financial in China transitioned from payments to digital credit provision. - Fintech firms in the United States grew to make up 38 percent of the unsecured personal loan market in 2018, from 5 percent in 2013 (TransUnion 2019). - In the United Kingdom and other markets, SMEs are a target: SMEs receive 2 percent of bank loans while contributing 50 percent of GDP and 70 percent of employment. - COVID-19 period behavior: - Some digital lenders have responded quickly to SME liquidity needs (examples cited: China and the United Kingdom) due to advantages in digital verification and onboarding. - Other fintech lenders halted new lending during lockdowns to preserve liquidity and manage credit risks. *Source: IMF staff.* ### part in the governments’ emergency lending programs, but the extent varies ### part in the governments’ emergency lending programs, but the extent varies ### Marketplace lending and fintech credit trends - Marketplace lending doubled from 2015 to 2017. - By 2017, marketplace lending reached US$400 billion. - Marketplace lending volume remains very small, at less than 0.5 percent of GDP for most countries. - In 2017, fintech credit was dominated by China, followed by the United States and the United Kingdom—together, they made up 98 percent of the fintech credit market. - Marketplace lending is likely to have declined substantially in China after 2017, following regulatory changes. - Marketplace lending is provided by a relatively large set of lenders and consists mostly of very small loans, likely to small borrowers (individuals and businesses). ### Digital financial inclusion indices, coverage, and trends - New indicators introduced: - “Digital” financial inclusion index: aggregates digital payment services provided through mobile phone and the internet (sample spans 2014–17). - “Traditional” financial inclusion index: constructed using the same approach for services provided by traditional financial institutions (sample spans 2011–17). - Sample covers 52 EMDEs. - Comprehensive financial inclusion index combines traditional and digital components via a three-stage principal component analysis: - Stage 1: compute “access” and “usage” for traditional and digital payment services. - Stage 2: compute traditional and digital financial inclusion indices from stage 1 outputs. - Stage 3: combine traditional and digital indices into a comprehensive financial inclusion index. - Digital financial inclusion has increased significantly during 2014–17, while traditional financial inclusion remained broadly unchanged in the same period. - In some countries digital financial inclusion appears to have been a game changer: - Comprehensive financial inclusion increased in most countries between 2014 and 2017. - Examples of greater progress in digital inclusion: Benin, Ghana, Senegal. - Examples of greater progress in traditional inclusion: Mongolia, Namibia, Peru. - In eight cases (including Zimbabwe, South Africa, and Nigeria) progress in financial inclusion is entirely driven by fintech—the increase in digital financial inclusion coincides with a fall in the traditional index. ### Regional patterns and within-region variation - Regional summary: - African and Asian countries maintain an overall lead in digital financial inclusion. - In Europe and Latin America, traditional financial inclusion dominates. - Considerable variation within regions: - In Africa, front runners in digital financial inclusion include Ghana, Kenya, and Uganda; laggards include Nigeria, Madagascar, Republic of the Congo. - Use of accounts (2017): - Middle Eastern countries in the sample tend to use almost exclusively accounts in financial institutions; mobile accounts are barely used. - Mobile money accounts are generally more present in African countries; in some African countries the share of adults with a mobile account is larger than the share with traditional accounts. ### Gender gaps and fintech’s role - Fintech is contributing toward closing financial inclusion gender gaps, with differences across regions. - Gender gaps tend to be slightly lower for fintech-driven financial inclusion than for traditional financial inclusion. - Regional variation: - Fintech plays a positive role in closing gender gaps in Middle East and African countries. - Gaps are lower for traditional financial inclusion in Asian and Latin American countries in the sample. - Variation across countries may be explained by obstacles fintech cannot address, such as cultural or social norms, and barriers in financial and digital literacy. ### Determinants of digital credit and marketplace lending - Marketplace lending fills a gap: it is higher in countries that have less financial depth. - Larger digital lending is associated with: - Better credit information, - Larger access to the internet, - Stronger legal rights. - Marketplace lending requires a sufficiently developed environment to thrive. ### COVID-19 implications and short- versus long-term effects - Long-term: COVID-19 has the potential to accelerate progress in digital financial inclusion (historical precedent: SARS epidemic in 2003 accelerated China’s launch of digital payments and e-commerce). - Anecdotal and early empirical evidence of acceleration: - Mobile money transactions in Rwanda increased by 450 percent between January and April 2020. - Number of users sending money virtually in Rwanda: 0.6 million in the week before lockdown, 1.2 million in the week after lockdown, and 1.8 million in the final week of April. - Download of finance mobile applications is estimated to have increased by 24 percent on average across 74 countries since the start of the lockdown, compared to prior trends. - Short-term risks: - The divide in progress in digital financial inclusion across and within countries could widen. - Developing digital infrastructure takes resources and time; countries with low access may struggle to scale up quickly given competing health and economic spending priorities. - Countries with already high access to digital financial services are likely to accelerate adoption further due to higher demand and supportive measures. - Short-term opposing impacts on demand for digital payments: - Increase in digital/contactless payments due to social distancing and e-commerce shifts. - Decrease in use of digital payments due to curtailed spending, fall in incomes, and employment losses. ### Methodological and data limitations - The index covers only 52 EMDEs due to lack of comparable data. - Raw data does not distinguish between digital services provided by fintech companies or banks; this precludes analysis of whether fintech companies are competing with or complementing services provided by traditional financial institutions. - Data sources include IMF Financial Access Survey, World Bank Global Findex Database, and GSMA (methodological details in Annex 2). *Source: pffiea - part in the governments’ emergency lending programs, but the extent varies* ### Box 1. Measuring Fintech-Driven (Digital) Financial Inclusion (continued)Box 1. Measuring Fintech-Driven (Digital) Finan ### Box 1. Measuring Fintech-Driven (Digital) Financial Inclusion (continued) ### Macroeconomic impact and empirical approach - Empirical analysis examines drivers of real GDP growth over 2014‒18 and 2011‒18 in a sample of 52 countries, using standard cross-country OLS regressions (Annex 4 and Khera and others 2020). - Initial values of traditional and digital financial inclusion are related to subsequent average growth, with a standard set of country-level control variables; initial values are used to reduce biases stemming from reverse causation. - Footnote caveat: “To establish a robust causal link between growth in digital financial inclusion and GDP growth, identifying a valid instrument for change in digital financial inclusion over the time span of GDP growth will help overcome potential biases stemming from endogeneity or reverse causation. This is explored more in detail in Khera and others (2020).” ### Key empirical findings on growth - Digital financial inclusion is significantly positively associated with growth, while traditional financial inclusion does not show the same positive association (Figure 11). - Possible interpretation: impact from traditional financial inclusion may have been reaped prior to the analysis period, whereas benefits of digital financial inclusion have only just started. - The positive association may reflect benefits from digital payments on transaction costs, liquidity, and creditworthiness (Islam, Muzi, and Rodriguez Meza 2018). - The analysis captures payments only and does not cover several components of digital finance (savings, credit, and insurance); therefore it is likely to underestimate the full impact on growth. - The impact of digital credit on sustainable growth depends on its ability to finance longer-term investment—which remains an open question. - Figure 11: annual GDP growth rates for countries with low (25th percentile), median, and high (75th percentile) levels of financial inclusion were calculated holding other explanatory factors at their median levels. Numeric values shown in the figure: 1.70, 2.30, 2.25, 2.20, 2.15, 2.10, 1.95, 1.90, 1.80, 1.85, 1.75, 2.05, 2.00 (95% confidence interval; source: IMF staff calculations). ### Role of fintech relative to traditional providers - Fintech companies targeting under- and unserved populations have had a limited disruptive impact on traditional bank operations so far; fintech services often complement, rather than substitute, traditional providers. - Typical fintech offerings to small clients include small loans of short duration, aggregators of services on digital platforms, and 24/7 access to online lending platforms that allow SMEs to seek financing outside business hours. - Fintech–bank collaboration is increasing: partnerships leverage banks’ regulatory and compliance experience and fintechs’ platforms for customer outreach; in some EMDEs fintech companies manage digital microcredit on behalf of banks. - Competition is emerging: purely digital banks and fintech lenders increasingly compete with informal money lenders, microfinance institutions, and small banks; big banks are acquiring or investing heavily in fintech. - Investment dynamics: banks’ combined investment in 2018 overtook venture capitalists’ investment in fintech companies (Figure 12 indicates bank tech spending in 2019 measured in US$ billion; note references to VC investment in US fintech and Europe fintech and VC data based on 2018 levels). ### Enabling factors for digital financial inclusion - Identified enablers include customer identification, digital infrastructure, financial literacy, and a supportive regulatory and legal environment. - Customer identification examples: - Telephone numbers used in EMDEs as identification for basic services such as payments. - Centralized databases for customer due diligence identification. - Work on “digital portable identity” with regulatory authorities (example: Financial Conduct Authority in the United Kingdom). - Aadhaar in India: a national biometric identification system issued to more than 1 billion people, used for national services and digital financial services and to satisfy regulatory requirements on customer identity. - Biometric identification introduced in developing Pacific countries such as Papua New Guinea and Samoa to allow unregistered persons to use fintech-based payments. - Regulatory/legal issue: need to balance information sharing and privacy protection. ### Constraints and risks highlighted by stakeholders - Two major fintech interviewee constraints: uncertainty of the regulatory environment and lack of technological expertise (“the coders”). - Uncertainty or frequent regulatory changes can be more constraining than a clear road map with tighter regulation. - In some countries, COVID-19 response measures channeled mainly through the banking sector could exacerbate fintech constraints. - Shortage of technological expertise is especially acute in EMDEs. - Funding constraints to scale-up, intensified during the COVID-19 crisis: - Initial funding often from incubators, accelerators, angel investors, and crowdfunding; later funding from private equity, venture capital, and hedge funds; some fintechs are publicly listed. - Fintech funding activity stalled in Q1 2020 across regions: Asia saw a 69 percent drop in funding and a 23 percent drop in deals quarter-over-quarter. - Regulatory authorities face challenges: catching up with a fast-changing landscape, budgetary constraints, lack of expertise, and lobbying pressures from traditional financial institutions. - Regulators and providers are adopting RegTech and SupTech to enhance regulatory processes and supervision. - Example of RegTech benefit: according to one RegTech company, cost of compliance for a client went down from £18 million to £0.5 million per annum by switching to their technology. - Demand-side constraint: lack of financial literacy or unfamiliarity with new technology, limiting outreach of COVID-19 related economic support to the most needy. - Fintech firms have added tutorials; Singaporean authorities have undertaken wide-ranging financial education initiatives. - Lack of access to e-government in several countries during COVID-19 is a major constraint to reaching the most needy. ### Empirical determinants of digital financial inclusion - Better access to digital infrastructure (availability of internet and mobile phones) is associated with higher usage of digital payments and credit; the relationship is monotonic and positive at all levels of digital infrastructure (Figure 13). - Increasing the number of mobile money agents in the same proportion would also lead to improvements in digital financial inclusion (magnitude smaller than for digital infrastructure). - The efficiency of traditional providers matters: more inefficient banking systems (with higher overhead costs to total assets) are associated with more digital financial inclusion. - Figure 13 displays the marginal effect of digital infrastructures and traditional financial familiarity on fintech usage indices for 2014–17 based on pooled regressions holding other explanatory factors at their mean level (source: IMF staff calculations; see Annexes 2 and 5). *Source: IMF staff calculations and interviews, as presented in the supplied content.* ### 1. Marginal Effects of Digital Infrastructure on Fintech Usage2. Marginal Effects of Financial Familiarity on Fintech Us ### 1. Marginal Effects of Digital Infrastructure on Fintech Usage2. Marginal Effects of Financial Familiarity on Fintech Us ### Factors that enable and constrain digital financial inclusion - Usage of fintech payment services is higher where there is already a high usage of traditional financial services; this could reflect higher financial literacy and trust in the financial system. - Institutions matter for development of mobile money agents; quality of governance is positively associated with availability of mobile money agents. - A more consumer-friendly environment (i.e., higher mobile money regulation index) is associated with greater adoption of mobile money. - On the credit side, availability of borrower information and higher protection of legal rights tend to support emergence and development of fintech credit (Annex 3). - Priorities to promote digital financial inclusion should depend on country circumstances: - Where traditional access is low, there is room to improve financial inclusion through fintech, irrespective of level of usage. - Where traditional usage is low, enhancing financial literacy and broader familiarity with financial services is essential to support financial inclusion, irrespective of access. - COVID-19 experience underscores importance of promoting digital services to the most needy; fiscal policy should include investment in digital infrastructure such as access to electricity, mobile, and internet coverage, digital ID among others (IMF 2020a). - In countries with higher digital access, the crisis could accelerate initiatives related to building conducive regulatory and institutional frameworks; efforts should be complemented by promotion of consumer and data protection, cyber-security, interoperability, and financial and digital literacy. ### Regulatory assessment and policy actions - Regulators have begun to assess fintech-related risks and formulate policies; these should be accelerated during and after the COVID-19 crisis. - Financial Stability Board (FSB) concluded that fintech and Big Tech do not yet present systemic risks (FSB 2017, 2019). - Past experience: push for financial inclusion without proper regulation contributed to the 2008 global financial crisis. - Digital lending raises concerns about predatory lending practices, which could become more prevalent during COVID-19 (Faux 2020). - Examples of enforcement: Indonesia’s Financial Services Authority identified and closed down more than 1,000 illegal peer-to-peer lenders offering prohibited financial services or operating without a proper license. - Sound policy approach at global and domestic levels is crucial (IMF 2019a); global cooperation needed to mitigate risks related to global monopolies, regulatory arbitrage, cross-border activities, cybersecurity, and money laundering (IMF 2019b, 2018a). - Domestic policy actions listed include: protecting data; preventing cyber risk; facilitating digital infrastructure; strengthening regulatory and supervisory frameworks; upgrading payment and securities settlement systems; ensuring standardization and interoperability; and developing effective user protection and contingency planning. - Supervisory approaches: adapt regulatory approaches to balance enabling innovation and addressing financial integrity, consumer protection, and financial stability; examples include innovation hubs and regulatory sandboxes. Regulation should remain proportionate to the risks and support safe use of innovative technologies (Taylor and others 2020). ### Risks of fintech to financial inclusion (direct) - Minimum investment requirements ("tech capital" and human capital) can exclude those without means, potentially creating new sources of financial exclusion among women, the poor, and the elderly in both EMDEs and advanced economies (G20 2019). - COVID-19–induced shift toward digital financial services could exacerbate exclusion of groups left behind. - Easy digital credit creates risks for people with limited financial literacy (Kaffenberger, Totolo, and Soursourian 2018). - Big data analytics and algorithmic profiling can both extend credit to those with limited credit history and entrench biases in historical data, potentially perpetuating unfair treatment and exclusion (Jagtiani and Lemieux 2017; FinRegLab 2019). - Economic shocks like COVID-19 will test reliability of existing models and indicators, potentially requiring adjustments and recalibrations. - FATF standard on ML/TF promotes a risk-based approach that encourages measures to meet national financial inclusion goals without undermining ML/TF measures; improper or disproportionate implementation (including via big data analytics) may aggravate financial exclusion (e.g., blanket exclusion of higher-risk customer categories). ### Risks of fintech to financial inclusion (procyclicality and indirect effects) - Fintech credit is small but growing rapidly; automation of credit decisions makes credit extension more frequent and faster. - Automation could lead to procyclicality if algorithms substitute for long-term client relationships and lead to faster contraction during downturns (Carstens 2018). - Tightening of funding conditions for fintech lenders—many of which are new and less established—could lead to sharper retrenchment, curtailing access to financial services for SMEs and low-income households disproportionately. - Potential consolidation could increase industry concentration with a few large firms emerging as dominant players. - Where fintech (and Big Tech) intermediate small deposits, banks’ funding structures may become more dependent on wholesale deposits, which could be more volatile and lead to credit contraction harmful to marginal borrowers. - Indirect risk: microfinance institutions and small banks may suffer competitive pressure from fintech; if they scale back before fintech sufficiently scales up, financial exclusion could increase. COVID-19 could exacerbate this as many microfinance institutions lack expertise and resources to expand digital operations in the near term. - Evidence of industry disruptions in some countries (for example, reported occurrences in Indonesia). ### Trust, user protection, and safeguarding customer funds - A loss of trust in digital technologies can set back progress in financial inclusion; increased availability of personal data can facilitate targeting but loopholes or fraud can erode trust. - Regulators and stakeholders noted value in a code-of-conduct directive for fintech firms, especially those dealing with retail customers. - Inadequate user protection risks undermining digital financial inclusion: - Mobile money operator bankruptcy or failure of banks holding pooled deposits could prevent full recovery of user balances. - Mitigation options include legal structures ensuring segregation of customer funds, investing customer funds in highly safe and liquid assets diversified across the safest banks, or requiring mobile money operators to hold customer funds as central bank reserves. - China, Peru, the Philippines, and Thailand require that the e-float be deposited at the central bank; Adrian and Mancini-Griffoli (2019) refer to this scheme as synthetic CBDC (sCBDC). ### Future agenda and international cooperation - International agreements are needed to address data privacy, cybersecurity, cross-border digital currencies, and digital identification; developing other standards will be difficult due to large differences across countries (e.g., national preferences regarding information sharing and data privacy). - International agreements also needed on anti-trust laws to ensure adequate competition in fintech and financial services; Big Tech firms (Alibaba, Amazon, Apple, Facebook, Google, Tencent) present competitive advantages and could put smaller companies out of business. - Concern that small countries and their regulatory policies could be captured by global giants. - Financial and digital literacy is scarce in both advanced economies and EMDEs; few countries mandate financial literacy courses in high school or college. Example: one emerging market introduced such a course as a high school graduation requirement but ran out of qualified teachers. - Authorities should undertake measures to increase financial and digital literacy, including creating incentives for private digital service providers to educate customers. - Macrofinancial risks: fintech increases interconnectedness between banks and (often unregulated) nonbanks, posing spillover risks; fintech could lead to “excessive” financial inclusion when access grows under insufficient regulation and supervision (examples cited: US subprime crisis; rise in default rate to nearly 20 percent on mobile bank loans in Kenya). - In crafting laws, ensure proportionality in regulation of small fintech firms (Adrian and Mancini-Griffoli 2019) while recognizing unsecured digital credit and light regulation raise complex crisis management issues. ### Role of fintech in COVID-19 response and recovery - Fintech has been useful in ensuring access to financial services and helping deliver governments’ support measures. - Fintech’s role in recovery depends on industry resilience to the shock and how the fintech landscape evolves post-COVID-19. - As more data become available, examine relationship between adoption of digital financial services and how well economies absorb the COVID-19 shock and recover post-COVID. ### Gender and political economy considerations - Women face multiple obstacles accessing finance (lower literacy/numeracy, lack of documentation, family responsibilities, social attitudes); fintech solutions appear well adapted to some constraints (consumer-friendly interfaces, no physical presence required). - AFI identified leveraging digital financial services as a top action point to address gender gaps in financial inclusion (AFI 2017). - A forthcoming IMF study is examining fintech’s evidence on bridging gender gaps (Khera and others 2020). - Political-economy question: should policymakers be concerned if new technology does not serve large segments of lower-income society even if GDP impact is large? Ignoring the “small guy” may carry political and social costs, as seen in social unrest during COVID-19. ### Evidence base and outreach - Interviews conducted with more than 50 companies active in payments, credit, insurance, and stock trading, covering incumbents and tech companies. - Interviews also conducted with authorities including: The Bank of England; Financial Innovation Bureau of the Korean Financial Services Commission; Luxembourg regulators; Malaysian regulator; Superintendency of Perú; central banks of China, Indonesia, Kenya, Morocco, Rwanda, Sierra Leone, and Uganda. - Outreach included meetings with Korea Fintech Association, National Payment Corporation of India (NPCI), Alliance for Financial Inclusion, NITI Aayog (India), former officials of the Reserve Bank of India, and the Gates Foundation. *Source: pffiea - 1. Marginal Effects of Digital Infrastructure on Fintech Usage2. Marginal Effects of Financial Familiarity on Fintech Us* ### Annex 1. Outreach and Interviews ### Annex 1. Outreach and Interviews ### Outreach: Firms, institutions, and fintech actors interviewed - Comprehensive list of companies and organizations interviewed (country and main activity preserved as in source), including but not limited to: - 1debit Inc. (“Chime”) — U.S. — Digital Bank - Activehours, Inc. (“Earnin”) — U.S. — Domestic Payday P2P Lending - Ant Financial Services Group — China — Online payment services provider - Aye Finance Private Limited — India — Nonbank digital credit provider - Bank Zero — South Africa — Digital bank - Barclays Bank Kenya — Kenya — Bank - Branch International, Inc. — U.S. — Digital credit provider to emerging markets - Cape Innovation and Technology Initiative — South Africa — Tech incubator - CapFloat Financial Services Private Limited — India — Nonbank digital credit provider - CBA Bank — Kenya — Bank - Cellulant — Kenya — Digital payments service provider - Click2Sure Intermediaries Proprietary Limited — South Africa — Digital insurance provider - Data Integrated — Kenya — Digital payments service provider - DIFC Fintech Hive — UAE — Fintech accelerator - First Community Bank — Kenya — Bank - Funding Circle Holdings Plc — U.K. — Digital credit provider to SMEs - Hippo Analytics Inc. (“Hippo Insurance”) — U.S. — Digital insurance provider - Honest Fund Co., Ltd — Korea — Marketplace lending platform - Indifi Technologies Pvt. Ltd — India — Digital credit provider to SMEs - Jambopay — Kenya — Digital payments service provider - Jumo World Limited — South Africa — Digital credit provider - Kakao Pay Corp. — Korea — Mobile payment and digital wallet service provider - KB Kookmin Bank — Korea — Bank - KCB Bank — Kenya — Bank - LendingClub Corporation — U.S. — P2P Lending - LendInvest Limited — U.K. — Online mortgage lender and investing platform - Mastarcard — U.S. — Payments service provider - MatchMovePay Pte. Ltd — Singapore — Digital payments provider for enterprises - M-DAQ Pte. Ltd — Singapore — Platform service provider. Allows to prices and trades exchange-traded products in a multitude of currencies - Microsoft Limited — U.K. — Technology and software company - MoneyMatch Sdn Bhd — Malaysia — Cross-border payments company - NISA finance (Pty) Limited — South Africa — Online invoice and purchase order financing platform - Nova Credit Inc. — U.S. — Credit Rating Conversion for U.S. immigrants - OCA (Open Capital Advisors) — Kenya — Financial services and strategy consulting firm - One Mobikwik Systems Private Limited — India — Digital payment services provider - Onfido Limited — U.K. — RegTech - Open Financial Technologies Private Limited — India — Digital bank - PayPal Holdings, Inc. — U.S. — Digital payments and credit provider - PayTabs — Saudi Arabia — Digital payments service provider - Paytm Mobile Solutions Private Limited — India — Payments services provider - People’s Fund Loan Co., Ltd — Korea — Marketplace lending platform - Poynt Co. — U.S. — Digital payments service provider - PrimeKeeper — Malaysia — Digital payments service provider - Razorpay Software Private Limited — India — Digital payments service provider - Samsung Electronics Co., Ltd — Korea — Largest Korean conglomerate, developer of Samsung’s fintech payments solution - SC Ventures — Singapore — Innovation, ventures and fintech investments unit of Standard Chartered Bank - Shinhan Bank Co., Ltd — Korea — Bank - Social Finance, Inc. — U.S. — Digital credit provider - Suade Labs Limited — U.K. — RegTech - TALA — Kenya — Digital credit provider - Thisisme — South Africa — Digital identity verification and due diligence platform - Tyme Bank Limited — South Africa — Digital bank - Uala — Argentina — Mobile personal financial management services provider - Uprise.Africa — South Africa — Equity crowdfunding platform - Viva Republica Co., Ltd (“Toss”) — Korea — Mobile financial services platform - Wazinsure — Kenya — Digital insurance provider - Woori Bank Developer Lab — Korea — Fintech startup accelerator - Yoco Technologies (Pty) Limited — South Africa — Point-of-sale payments provider for small businesses - Zerodha Commodities Private Limited — India — Online stock trading platform ### Methodology: Constructing financial inclusion indices (Annex 2) - Coverage: 52 developing and emerging economies, years 2014 and 2017. - Approach: Three-stage principal component analysis (PCA). - Stage 1: Construct sub-indices for Access and Usage separately for traditional (FI_Ta, FI_Tu) and fintech-driven components (FI_Fa, FI_Fu) based on selected variables (Annex Table 1). Weights for each variable are in Annex Figure 1. - Stage 2: Combine Access and Usage indices (separately for traditional and digital) via PCA to obtain overall traditional and fintech-driven financial inclusion indices. Notation: - (FI_T)it = β1 (FI_Ta)it + β2 (FI_Tu)it + eit - (FI_F)it = α1 (FI_Fa)it + α2 (FI_Fu)it + μit - If the model is well specified, E(e) = 0 and E(μ) = 0. - Stage 3: Apply PCA on the two indices to compute comprehensive financial inclusion index: - FIit = ω1 (FI_T)it + ω2 (FI_F)it + ωit - Data treatment: - Winsorize each variable at 2nd and 98th percentile to avoid extreme values. - Normalize all indices between 0 and 1 across countries and years; value of 1 indicates the country-year with the highest level of financial inclusion among countries and years. - Annex Table 1: Selected variables and weights for Access and Usage in Traditional and Digital indices (weights preserved in table). - Examples of variable-weight pairs preserved as in source: - Traditional Access: Number of ATMs per 100,000 adults (IMF FAS) — Weight 0.25 - Traditional Usage: % of adults with a financial institution account (WB Findex) — Weight 0.25 - Digital Access: Mobile subscription per 100 people (ITU) — Weight 0.125 - Digital Usage: % of adults who have a mobile account (WB Findex) — Weight 0.125 - Annex Figure 1: First-stage PCA weights for individual variables across Access and Usage (values preserved as presented in figure). ### PCA loadings and variance explained (Annex Tables 2–3) - Annex Table 2: Principal Component Analysis: Loadings (PC1 shown) - Access (Traditional) PC1: - ATM per 100,000 population: 0.7071 - bank per 100,000 population: 0.7071 - Usage (Traditional) PC1: - Account at an F.I (%): 0.4842 - Saving at an F.I (%): 0.3954 - Debit Card (%): 0.4820 - F.I account for wages (%): 0.4551 - F.I account for utility (%): 0.4120 - Access (Fintech-driven) PC1: - Electricity (%): 0.7071 - Internet (%): 0.7071 - Usage (Fintech-driven) PC1: - Mobile account (%): 0.5130 - Use internet to pay (%): 0.3722 - Mobile for wages (%): 0.5356 - Mobile for utility (%): 0.5580 - Annex Table 3: Cumulative Variance Explained by Principal Components (values preserved) - Access (Traditional): PC1 0.7982, PC2 1.0000 - Access (Fintech-driven): PC1 0.7884, PC2 1.000 - Usage (Traditional): PC1 0.7759, PC2 0.8986, PC3 0.9623, PC4 0.9849, PC5 1.0000 - Usage (Fintech-driven): PC1 0.7495, PC2 0.9311, PC3 0.9774, PC4 1.0000 - Access and Usage (Fintech-driven): PC1 0.5435, PC2 1.0000 - Traditional Financial Inclusion Index: PC1 0.8448, PC2 1.0000 - Fintech-driven Financial Inclusion Index: PC1 0.5000, PC2 1.0000 - Overall Financial Inclusion Index: PC1 0.6083, PC2 1.0000 - Note: For the fintech-driven index, mobile money agents could not be combined with other fintech access indicators in first-stage PCA due to negative correlation with internet and mobile subscription; mobile money agents were added at the second stage. ### Determinants of fintech (marketplace) lending — hypotheses and empirical approach (Annex 3) - Research questions: - Role of basic infrastructure—technology, information, legal—in marketplace lending development. - Whether digital credit develops in countries with more developed financial sectors or complements lack of access to finance. - Whether marketplace lending overcomes geographical barriers in access to credit. - Whether features of the banking sector matter for marketplace lending. - Hypotheses (preserved as in source): - Economic development: Higher income per capita expected to increase demand for marketplace lending. - Internet: marketplace lending positively related to share of population using internet. - Information: marketplace lending positively related to easier and wider access to credit information through credit registries. - Legal infrastructure: marketplace lending positively related to strength of legal rights. - Financial development (depth): marketplace lending negatively related to depth of credit provided by traditional lenders (banks). - Financial development (access): lending typically positively related to financial access but measure available relates to banks and ATMs and doesn’t include digital access. - Financial development (efficiency): relationship ambiguous. - Geographical barriers: marketplace lending negatively related to geographical presence of bank branches. - Features of Banking Sector (structure, profitability, asset quality): relationship ambiguous. - Country sample and period: panel of 105 countries during 2015–17. - Data source: Cambridge Centre for Alternative Finance (CCAF) for marketplace lending data. - Regression approach: - Fixed-effect panel regression with country and year fixed effects: - Yit = βi + γt + ϕ Xit + εit - Yit is logarithm of size of marketplace lending as share of nominal GDP (separate regressions for total, business, consumer fintech credit). - Xit includes: PPP GDP per capita, internet users, depth of credit information, strength of legal rights, three financial development sub-indices (depth, access, efficiency), bank branches per 1,000 square kilometers, and banking sector features (concentration, return on assets, share of nonperforming loans relative to gross loans). - Results: Annex Table 4 presents panel regression estimates for total fintech lending (sample: 2015–17). Key reported coefficients across models (exact values preserved from table): - Log (PPP GDP per capita): - Model (1): 7.113** - (2): 8.007** - (3): 8.054** - (4): 9.026*** - (5): 8.908*** - (6): 13.46** - (7): 6.119** - Internet users (% of population): - (1): 3.541*** - (2): 3.539*** - (3): 3.786*** - (4): 3.881*** - (5): 5.299*** - (6): 4.099*** - Depth of Credit Information: - (1): 0.251 - (2): 0.247 - (3): 0.173 - (4): 21.086 - (5): 0.466*** - Strength of Legal Rights: - (1): 0.298 - (2): 0.329 - (3): 0.322 - (4): 4.742*** - (5): 0.762 - Financial Development of Institutions – Depth: - (1): 23.363*** - (2): 23.528*** - (3): 24.893*** - (4): 23.379*** - Financial Development of Institutions – Access: - (1): 0.326 - (2): 0.584 - (3): 2.628 - (4): 1.124 - Financial Development of Institutions – Efficiency: - (1): 20.0237 - (2): 20.0379 - Number of Bank Branches (per 1000 km2): - (3): 20.692 - Bank Concentration (top 5 banks): - (4): 20.295 - Return on Assets of Banks: - (6): 0.0444 - Non-Performing Loans of Banks: - (7): 20.784* - Constant terms (preserved per model): - (1): 22.569*** - (2): 23.782*** - (3): 23.744*** - (4): 23.414*** - (5): 24.434*** - (6): 210.07*** - (7): 24.175*** - Sample sizes and fit statistics (preserved): - Observations: 210, 209, 206, 205, 189, 102, 180 - R-squared: 0.4630, 0.5420, 0.5460, 0.5800, 0.6300, 0.5890, 0.634 - Number of Countries: 105, 104, 103, 102, 92, 58, 83 - Notes from table (preserved): - All explanatory variables are standardized. - Standard Newey West standard errors used; baseline does not cluster standard errors by country or time. Robustness check clustering by country produced almost identical results. - Robust standard errors in parentheses. *** p , 0.01, ** p , 0.05, * p , 0.1. ### Digital financial inclusion and growth (Annex 4) - Research question: Does greater fintech-driven (digital) financial inclusion in payments translate into higher economic growth? How does its impact differ from traditional financial inclusion? - Methodology: - Cross-country OLS relating indices of financial inclusion (traditional and digital) at one point in time to subsequent growth over periods 2011–18 and 2014–18. - Baseline estimation: - ẏ(i)2011−18 = β0 + β1 FI(i)T,2011 + β2 FI(i)F,2011 + β3 X(i)2011 + ε(i) - ẏ is real GDP growth for 52 developing and emerging economies; initial financial inclusion indices used to mitigate reverse causality. - Control variables (X(i)2011) include: - Log of GDP per capita (IMF WEO) - Government consumption (% of GDP) (World Bank) - Foreign direct investment (% of GDP) (IMF WEO) - Inflation: annual % change in consumer prices (IMF WEO) - Level of financial depth: log of private credit (% of GDP) (World Bank) - Regional dummy variables: Asia, Middle East and Central Asia, Latin America, emerging Europe, sub-Saharan Africa - Robustness checks and additional analyses (preserved as in source): - Adding fintech credit: include log of total new originations of fintech credit as a share of nominal GDP in 2015 to regression for 2014–18; estimates indicate fintech credit effect on growth not statistically significant, while other variables remain similar to baseline. - Removing cyclical effects in GDP: replace dependent variable with average of detrended real GDP growth using Hodrick-Prescott (HP) filter. - Using comprehensive financial inclusion instead of separate indices: initial comprehensive FI has positive impact on per capita GDP growth, but not statistically significant once traditional financial sector depth (domestic private credit-to-GDP) is added. - Using Financial Development Index (FDI) instead of private credit (% of GDP). - Using three-year averages (2009–11 and 2011–14) for government consumption, FDI, and private credit (% GDP) instead of initial period only. - Overall: Robustness checks produced results similar to baseline estimates. For more detailed discussion, refer to Khera and others (2020). *Source: Annex 1–4 content as provided in the IMF PDF "pffiea - Annex 1. Outreach and Interviews."* ### Annex Table 5. Results: Growth and Digital Financial Inclusion ### Annex Table 5. Results: Growth and Digital Financial Inclusion ### Empirical setup and question - Question/motivation: What drives the adoption of fintech-driven financial inclusion in payments? Why do some countries have higher adoption rates than others? Is fintech-driven financial inclusion complementing or substituting traditional means of financial inclusion? - Estimation approaches: - Ordinary least squares (OLS) regression: drivers of access (supply) — dependent variable: mobile money agents (per 100,000 adults). - Fractional logit regression (Papke and Wooldridge 1996): drivers of fintech usage (demand) — dependent variable: fintech usage index (falls in [0,1]). - Equations estimated for 2014 and 2017, and pooled. Reported results refer to the pooled regression. Sample: 52 low-income and developing economies for which the financial inclusion index is computed. - To account for two-way relation between access and usage, lagged values of the fintech usage index and fintech access index are used as determinants to mitigate endogeneity. - Other determinants (lagged one-period): log of GDP per capita; overhead cost to assets (financial sector efficiency); bank concentration (assets of three largest commercial banks / total commercial banking assets); log of NPL as a share of total gross loans (financial stability); rule of law (scaled 0–1); urban population share. ### OLS estimation: Real GDP growth rate (2011–18) and (2014–18) — key coefficient patterns (Annex Table 5) - Log (GDP per capita): - (1) 0.0879 *** - (2) 0.0940 *** - (3) 0.0880 *** - (4) 0.0933 *** - (5) 0.0837 *** - (6) 0.0793 *** - (7) 0.0465 *** - (8) 0.0439 *** - (9) 0.0447 *** - (10) 0.0474 *** - (11) 0.0473 *** - (12) 0.0419 *** - Inflation: - (1) 0.0035 - (2) 0.0085 ** - (3) 0.0091 ** - (4) 0.0075 * - (5) 0.0080 * - (6) 0.0099 ** - (7) 0.0015 - (8) 0.0039 - (9) 0.0036 - (10) 0.0029 - (11) 0.0030 - (12) 0.0018 - Traditional F.I Index: - (1) 0.157 - (2) 0.241 ** - (3) 0.176 - (4) 0.129 - (5) 0.004 - (6) 0.144 - (7) 0.0606 - (8) 0.109 - (9) 0.112 - (10) 0.024 - (11) 0.023 - (12) 0.105 - Digital F.I Index: - (1) 0.272 * - (2) 0.364 *** - (3) 0.376 *** - (4) 0.311 ** - (5) 0.314 ** - (6) 0.371 *** - (7) 0.175 ** - (8) 0.150 ** - (9) 0.148 ** - (10) 0.129 * - (11) 0.130 * - (12) 0.188 ** - Gov. Consumption (% of GDP): - (1) 0.016 *** - (2) 0.016 *** - (3) 0.016 *** - (4) 0.016 *** - (5) 0.012 *** - (6) 0.0096 *** - (7) 0.0096 *** - (8) 0.0091 *** - (9) 0.0091 *** - (10) 0.0048 * - FDI (% of GDP): - (1) 0.0025 - (2) 0.0038 * - (3) 0.0046 ** - (4) 0.0007 - (5) 0.0002 - (6) 0.0001 - Log (private credit) (% of GDP): - (1) 0.0401 ** - (2) 0.0504 ** - (3) 0.0488 * - (4) 0.0357 ** - (5) 0.0359 ** - (6) 0.0347 * - Constant: - (1) 0.929 *** - (2) 1.182 *** - (3) 1.141 *** - (4) 1.064 *** - (5) 0.970 *** - (6) 0.950 *** - (7) 0.468 *** - (8) 0.578 *** - (9) 0.584 *** - (10) 0.502 *** - (11) 0.500 *** - (12) 0.470 *** - Regional dummy included in columns: (6) Yes, (12) Yes. Other columns: No. - Sample sizes (N): - (1) 48; (2) 48; (3) 48; (4) 47; (5) 47; (6) 47; (7) 50; (8) 50; (9) 50; (10) 49; (11) 49; (12) 49. - Adjusted R-squared (adj. R2): - (1) 0.238 - (2) 0.442 - (3) 0.451 - (4) 0.476 - (5) 0.509 - (6) 0.543 - (7) 0.154 - (8) 0.370 - (9) 0.356 - (10) 0.432 - (11) 0.418 - (12) 0.486 ### Interpretation of Table 5 (key empirical findings) - The Digital F.I Index shows positive and often statistically significant coefficients across pooled specifications (ranging from 0.129* to 0.376***), indicating a positive association between digital financial inclusion and real GDP growth in these OLS specifications. - Traditional F.I Index coefficients are smaller and less consistently significant across specifications. - Log (GDP per capita) is consistently positive and highly significant across all columns. - Government consumption (% of GDP) shows positive and frequently significant coefficients. - Financial depth proxied by log(private credit) (% of GDP) is positive and significant in several specifications. - Model fit (adj. R2) improves in some specifications when additional controls or regional dummies are included (adj. R2 up to 0.543 in column (6)). *Note: Robust standard errors are in parentheses in original table. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1.* --- ### Annex 5. Determinants of Digital Financial Inclusion ### Methodology summary (Annex 4 / Annex 5) - Drivers of fintech adoption analyzed separately for access (supply) and usage (demand) using: - OLS for access (mobile money agents per 100,000 adults). - Fractional logit for usage (fintech usage index in [0,1]). - Lagged explanatory variables used to reduce reverse causality. - Determinants included (lagged): log GDP per capita; overhead cost to assets; bank concentration; log NPL ratio; rule of law; urban population share; lagged traditional access and usage indices; lagged fintech access/usage indices; mobile money agents (lagged). ### Determinants of Fintech Usage — selected coefficients (Annex Table 6, Fractional Logit Estimation) - Trad. Access F.I Index (t−3): - (1) 3.658 *** - (2) −2.418 ** - (3) 4.266 *** - (4) 4.181 *** - (5) 4.205 *** - (6) 4.006 *** - (7) 4.008 *** - (8) 3.909 *** - (9) 2.030 * - (10) 1.912 * - (11) 1.980 * - (12) 1.996 * - (13) 2.300 ** - (14) 2.228 ** - Trad. Usage F.I Index (t−3): - (1) 4.692 *** - (2) 4.449 *** - (3) 5.092 *** - (4) 5.388 *** - (5) 5.654 *** - (6) 5.660 *** - (7) 5.346 *** - (8) 5.354 *** - (9) 4.956 *** - (10) 5.199 *** - (11) 5.311 *** - (12) 5.322 *** - (13) 4.935 *** - (14) 4.882 *** - Fintech Access F.I Index (t−3): - (1) 0.211 - (2) 1.236 - (3) 1.165 - (4) 1.425 * - (5) 1.814 ** - (6) 0.374 - (7) 0.607 - (8) 1.354 * - (9) 1.534 ** - (10) 1.669 ** - (11) 1.658 ** - (12) −0.272 - (13) −0.229 - Mobile Money Agents (t−1): - (1) 0.00302 *** - (2) 0.00201 *** - (3) 0.00170 *** - (4) 0.00184 *** - (5) 0.00169 *** - (6) 0.00114 * - (7) 0.00107 * - (8) 0.00264 *** - (9) 0.00235 *** - (10) 0.00248 *** - (11) 0.00249 *** - (12) 0.00170 ** - (13) 0.00166 ** - Log (NPL) (t−1): - (1) 0.501 ** - (2) 0.470 ** - (3) 0.511 ** - (4) 0.481 ** - (5) 0.330 - (6) 0.320 - Log (GDP Per capita): - (1) −0.538 ** - (2) −0.578 ** - (3) −0.563 ** - (4) −0.571 ** - (5) −0.135 - (6) −0.0975 - Overhead cost To assets (t−1): - (1) 0.0979 - (2) 0.0923 - (3) 0.122 * - (4) 0.112 - (5) 0.127 * - (6) 0.0898 - (7) 0.0808 - (8) 0.0794 - (9) 0.102 - (10) 0.108 * - Rule of law (t−1): - (1) −0.0122 - (2) −0.0135 - (3) 0.000135 - (4) −0.000901 - (5) −0.00716 - (6) −0.00710 - (7) 0.00307 - (8) 0.00284 - Urban Population (t−1): - (1) −0.00780 - (2) −0.00410 - (3) 0.000578 - (4) −0.00276 - Year fixed effect included in columns: (7), (8), (13), (14). Sample sizes N vary by column (96, 80, 79, 95, etc.). - Note: Robust standard errors reported in original. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1. ### Determinants of Fintech Access — selected coefficients (Annex Table 7, OLS Estimation: Mobile money agents per 100,000 adults (t)) - Trad. Access F.I. Index (t−3): - (1) −539.2 *** - (2) −374.0 * - (3) −358.9 * - (4) −346.9 * - (5) −207.7 - (6) −226.1 - (7) −203.0 - (8) −224.2 - (9) −72.56 - (10) −48.74 - (11) 46.53 - (12) 104.9 - (13) 104.8 - (14) 99.74 - (15) 99.08 - Trad. Usage F.I. Index (t−3): - (1) 119.8 - (2) 22.13 - (3) 82.98 - (4) 93.75 - (5) 166.4 - (6) 52.00 - (7) 85.29 - (8) −61.68 - (9) 105.4 - (10) 156.5 - (11) 166.7 - (12) 73.45 - (13) 72.18 - (14) −24.69 - (15) −33.33 - Fintech Usage F.I. Index (t−3): - (1) 396.4 ** - (2) 307.5 - (3) 225.7 - (4) 131.1 - (5) 102.7 - (6) 179.9 - (7) 153.0 - (8) 408.9 ** - (9) 321.3 - (10) 250.9 - (11) 189.2 - (12) 185.5 - (13) 244.8 - (14) 227.4 - Log (NPL) (t−1): - (1) 94.18 *** - (2) 83.67 *** - (3) 70.26 ** - (4) 51.97 - (5) 39.97 - (6) 40.78 - (7) 24.98 - Log (GDP per capita) (t−1): - (1) −109.0 *** - (2) −102.2 *** - (3) −113.9 *** - (4) −146.8 *** - (5) −143.9 *** - (6) −131.2 *** - (7) −116.0 ** - Overhead cost to assets (t−1): - (1) 28.82 ** - (2) 43.95 *** - (3) 51.91 *** - (4) 54.16 *** - (5) 51.94 *** - (6) 54.60 *** - (7) 24.95 ** - (8) 33.15 *** - (9) 41.67 *** - (10) 41.90 *** - (11) 41.87 *** - (12) 43.03 *** - Bank concentration (t−1): - (1) −3.817 ** - (2) −3.273 ** - (3) −3.149 ** - (4) −3.667 ** - (5) −3.576 ** - (6) −3.151 ** - (7) −3.183 *** - (8) −3.133 ** - (9) −3.193 *** - (10) −2.938 ** - Urban population (t−1): - (1) −2.665 * - (2) −3.013 ** - (3) −2.462 * - (4) −2.846 ** - (5) 6.235 *** - (6) 6.233 *** - (7) 6.275 *** - (8) 6.265 *** - Rule of law (t−1): - (1) 3.826 * - (2) 4.531 ** - (3) 3.826 * - (4) 4.531 ** - Year fixed effect included in columns: (7), (8), (15), (16) (as labeled in original table). Sample sizes N vary by column (96, 80, 79, 95, etc.). - Adj. R2 examples: - (1) 0.102 - (2) 0.256 - (3) 0.303 - (4) 0.354 - (5) 0.376 - (6) 0.397 - (7) 0.408 - (8) 0.441 - (15) 0.428 - Note: Robust standard errors reported in original. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1. ### Interpretation of Annex Tables 6 and 7 (key empirical findings) - Fintech usage is strongly and positively associated with lagged traditional usage F.I. Index (large positive coefficients, consistently statistically significant). - Mobile money agent density (lagged) has a positive and significant association with fintech usage (coefficients in the 0.00166–0.00302 range, many statistically significant). - Fintech access (mobile money agents per 100,000 adults) is positively associated with fintech usage (e.g., 396.4 ** in column (1) of Table 7) in some specifications, indicating complementarity between access and usage. - Higher log(NPL) is positively associated with fintech access in many specifications (e.g., 94.18 ***), suggesting fintech access may expand where traditional financial stability indicators show higher NPL ratios. - Overhead cost to assets is positively associated with fintech access in many specifications (e.g., 51.91 ***, 54.16 ***), consistent with the idea that higher banking inefficiency is linked to greater fintech agent proliferation. - Bank concentration shows negative association with fintech access (coefficients around −3.1 to −3.8, frequently significant), indicating more concentrated banking sectors are associated with fewer mobile money agents per 100,000 adults in several specifications. - Rule of law shows mixed but occasionally positive associations with fintech access (e.g., 3.826 *, 4.531 **). - Urban population effects vary across specifications: some negative coefficients in early columns and strong positive coefficients in later columns (e.g., 6.235 ***), indicating specification sensitivity. *Note: Robust standard errors are in parentheses in original tables. Significance: *** p , 0.01, ** p , 0.05, * p , 0.1.* _Italic: Source: IMF staff calculations (tables and notes as presented in the source content)._ --- _Source: https://www.imf.org/-/media/files/publications/dp/2020/english/pffiea.pdf_