## wpiea2022080-print-pdf - Executive Summary

## Source details

**Canonical URL:** [wpiea2022080-print-pdf - Executive Summary](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022080-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022080-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022080-print-pdf.pdf.json)

---

### Executive summary: core questions and conclusions
- Research questions:
  - Does Fintech improve digital financial inclusion?
  - Are there segments of society not included because they do not have the capacity and means to adopt Fintech (women, the poor, people in rural areas)?
- Main empirical conclusions (based on panel regression):
  - Fintech has a positive correlation with financial inclusion, with the largest correlation occurring when digital financial inclusion is used.
  - Fintech has a negative association with the class divide and rich-poor divide (i.e., it has the potential to make it better).
  - Fintech has no impact on the gender digital gap.
- Policy implication highlighted:
  - Fintech alone may not be sufficient to close the gender gap in access to financial services; Fintech development may need to be complemented with targeted policy initiatives aimed at addressing the gender gap directly, and at changing attitudes and social norms across demographic groups.

### Introduction: framing and measurement challenges
- Observations:
  - Fintech is transforming financial services by leveraging technology and cloud-based data to offer products better tailored to consumers’ needs at lower cost.
  - Few studies provide direct empirical evidence of Fintech’s impact on financial inclusion due to challenges measuring Fintech and the “digital” aspect of financial inclusion, patchy data, and short time series.
- Measurement approach used in this paper:
  - Use of new digital financial inclusion indicators and various Fintech proxies, specifically: Digital Financial Inclusion Index (Sahay et al. 2020), Fletcher School’s Digital Intelligence Index (DII), and World Bank’s Digital Adoption Index (DAI).

### Recent trends, literature, and the “dark side” of Fintech
- Scope of Fintech (categories identified):
  - (i) credit, deposits, and capital-raising services;
  - (ii) payments, clearing and settlement services, including digital currencies;
  - (iii) investment management services;
  - (iv) insurance.
- Empirical and regional patterns:
  - Significant increase in mobile accounts and mobile money usage between 2014 and 2017.
  - Digital financial inclusion increased between 2014 and 2017 even where traditional inclusion stalled or declined.
  - In eight countries (including Zimbabwe, South Africa, Nigeria) improvements in financial inclusion were entirely driven by Fintech (Sahay et al. 2020).
- Divergent findings on gender effects:
  - Sahay et al. (2020): digital financial inclusion gender gaps are lower on average than traditional financial inclusion gender gaps overall and in Africa, the Middle East and Central Asia; but higher in Asia-Pacific and LAC.
  - Chen et al. (2021): persistent “Fintech gender gap”; gender gap smaller for products that complement traditional banking.
  - BIS working paper: 29% of men use Fintech products and services versus 21% of women; Fintech gender gap 8 percent point.
- The “dark side” of Fintech:
  - Exclusion of women, the aged, the poor, and minority groups.
  - Algorithm biases and predatory lending practices.
  - Risk that digital delivery excludes those without access to digital infrastructure, financial/digital literacy, or trust.

### Mind the gap: measured divides and regional magnitudes
- Gender gap:
  - The gender gap exists across all regions and across measures (accounts, used internet to pay bills, made or received digital payments).
  - Average gender gap across all regions, using the accounts measure, is 11 percent points.
  - Example: in MEA the gender gap as measured by number of accounts is 19 percent points.
- Rich-poor (class) gap:
  - The average rich-poor gap is 15 percent points across all regions.
  - Regional magnitudes using accounts measure: LAC 21 percent points, EAP 19 percent points.
  - In SSA, MEA and SAS the rich-poor gap is smaller and narrows when digital measures are used.
- Other descriptive facts:
  - In the US 93% of adults have a bank account; there exists a 13 percent point gap between the wealthiest households and the poorest.
  - Internet access disparity cited: 48% of women have access to the internet compared to 55% for men globally (Alliance for affordable internet, 2021).

### Data, indices, and empirical approach
- Fintech proxies used (four):
  - Leapfrog (proxied by mobile phones/fixed line subscriptions)
  - Venture capital raised for Fintech companies (as % of GDP)
  - Total amount of funds raised through P2P lending platforms (as % of GDP)
  - Mobile money accounts
- Financial inclusion indicators used:
  - Digital Financial Inclusion Index (Sahay et al. 2020) — covers 52 EMDEs for 2014 and 2017; three-stage principal component analysis creating access and usage measures, and comprehensive financial inclusion index.
  - Fletcher School’s Digital Intelligence Index (DII) — covers 90 economies from 2008 to 2019 using 160 indicators across supply, demand, institutional environment, and innovation and change.
  - World Bank’s Digital Adoption Index (DAI) — composite index covering 171 countries for 2014–2016 using 16 indicators across businesses, people and governments.
- Measures of digital divide constructed:
  - Gender divide: deviation of female digital usage from male (higher is better = less divide).
  - Rural divide: deviation of rural digital usage from the average (higher is better).
  - Class divide: deviation of the bottom 40% from the top 60%.
- Summary statistics for indices:
  - Traditional Financial Inclusion Index — Mean 0.33; Standard deviation 0.22; Min-Max 0.0-1.0
  - Digital Financial Inclusion Index — Mean 0.36; Standard deviation 0.20; Min-Max 0.0-1.0
  - Digital Intelligence Index — Mean 0.51; Standard deviation 0.19; Min-Max 0.1-1.0
  - WB Digital Adoption Index — Mean 0.5; Standard deviation 0.19; Min-Max 0.1-0.8
- Correlation excerpts and interpretation:
  - Highest correlation is between the Digital Intelligence Index and the World Bank Digital Adoption Index (X 0.88).
  - Lowest correlation (0.2) is between the IMF traditional Financial Inclusion Index and the IMF-Fintech Driven Index.
  - The IMF-Fintech Driven Index has only a 0.3 and 0.4 correlation with the Digital Intelligence Index and the Digital Adoption Index respectively.
  - Correlations indicate Fintech may substitute for bank branches (Fintech financial inclusion indicator negatively correlated with branches).

### Empirical approach (model specification)
- Fixed-effect panel with robust standard errors:
  - FI_it = c + β1 X_it + β2 Fintech_it + α_i + ε_it  (equation (1))
  - XDIVIDE_it = β1 + β2 X_it + β3 Fintech_it + α_i + ε_it  (equation (2))
- FI is either digital or traditional financial inclusion; controls include per capita income, credit to private sector, education level, regulatory quality, population, internet penetration, and mobile phone penetration.
- Three digital divides (Fletcher School): gender divide, class divide, rural divide. Divide variables reversed in regressions so that a lower number (i.e. a smaller gap) means an improvement.
- Fintech proxies: Leapfrog (ratio of mobile phones to fixed lines), Capital Invested (venture capital raised for Fintech), Capital raised (P2P lending as percent of GDP), Business Volume (digital lending/capital raising).

### Does Fintech promote financial inclusion? — Key empirical results (Table 3 highlights)
- Fintech proxied by leapfrog and capital raised:
  - Positive and statistically significant in most regressions with digital financial inclusion and the Digital Intelligence Index (DII).
  - Leapfrogging has no significant effect with traditional financial inclusion (bank branches); capital raised has no correlation with the traditional financial inclusion indicator (Eq 2, 5, 8).
  - Leapfrogging shows negative correlation with traditional indicators and branches (implying Fintech is a substitute for bank branches).
- Selected reported coefficients (exact values as presented):
  - Population: 2.551*** (Digital FI, col 1); -5.835 (Branch, col 2)
  - Regulatory Quality: -0.077 (col 1); 0.056*** (DII, col 4)
  - Credit/GDP: -0.384 (col 1); 8.262*** (Branch, col 2)
  - Education (secondary): 0.011* (col 1); -0.112*** (Branch, col 2)
  - Cellphone/population: 0.003*** (col 1)
  - Internet/population: 0.074* (col 1); 0.149*** (Digital FI, col 3)
  - Leapfrogging (cell/fixed line): 0.037*** (Digital FI, col 1); -0.172** (Branch, col 2)
  - Fintech (Capital Invested): 0.013* (Digital FI, col 3); 0.004*** (DII, col 4)
  - Fintech (Capital raised, P2P): -0.014 (Digital FI, col 6); 0.004*** (DII, col 7); -0.206** (Branch, col 8)
- Model fit and sample sizes (selected):
  - Observations: 64 (col 1), 950 (col 2), 33 (col 3), 369 (col 4)
  - R-squared: 0.786 (col 1), 0.197 (col 2), 0.910 (col 3), 0.692 (col 4)
  - Number of ctry: 36 (col 1), 152 (col 2), 19 (col 3), 65 (col 4)
- Interpretation:
  - Leapfrog variable has the highest positive effect on the Digital Financial Inclusion Index and a negative association with branches — evidence of technological leapfrogging substituting for branch infrastructure.
  - Capital invested significant at 10% for digital financial inclusion (Eq 3).
  - P2P lending (capital raised as % of GDP) negatively affects branches, positively correlates with DII (Eq 7), but no significant relationship with the Digital Financial Inclusion Index (Eq 6) — implying complementarity with traditional bank lending but substitution for branches, and helping fill the lending gap for low-income households and SMEs.

### Fintech and the rural, class, and gender divides — Key results (Tables 4–6 highlights)
- Estimation framework controls for trust index, regulatory quality, education; gender divide regressions also include gender inequality index.
- Main findings:
  - Fintech proxies are significant and negatively correlated with the rural divide and the class divide — implying Fintech can improve rural and class divides.
  - Fintech proxies show no significant correlation with the gender divide.
- Rural Divide (selected coefficients and diagnostics):
  - Trust index: -0.382*** (col 1)
  - Leapfrog: -3.039* (col 2)
  - Digital financial inclusion: -4.948** (col 3)
  - Digital adoption index: -7.138** (col 4)
  - Fintech (Capital invested): -0.094** (col 5)
  - Fintech (Capital raised, P2P): -0.069** (col 5)
  - Observations: 147 (col 1); R-squared: 0.286 (col 1)
- Class Divide (selected coefficients and diagnostics):
  - Trust index: -0.575*** (col 1)
  - Leapfrog: -2.357*** (col 2)
  - Digital financial inclusion: -3.857*** (col 3)
  - Digital adoption index: -6.282*** (col 4)
  - Fintech (Capital invested): -0.056*** (col 5)
  - Fintech (Capital raised, P2P): -0.060*** (col 5)
  - Observations: 147 (col 1); R-squared: 0.450 (col 1)
- Gender Digital Divide (selected coefficients and diagnostics):
  - Regulatory quality: -0.416* (col 1); -0.443** (col 2); -0.416** (col 3)
  - Gender inequality: 4.555 (col 1); 3.553 (col 2); 1.561 (col 3)
  - Digital adoption index: -0.882 (col 2)
  - Fintech (Capital invested): 0.013 (col 2)
  - Fintech (Capital raised, P2P): -0.004 (col 3)
  - Observations: 41 (col 1); 137 (col 2); 98 (col 3); R-squared: 0.325 (col 1)
- Robustness:
  - Results using raw gap measures (e.g., digital payments female – digital payments male) yield similar conclusions (Appendix C).
- Caveat:
  - Explanatory powers of these variables are rather low; further analysis with more variables is suggested as new data become available.

### Mechanisms and examples
- Mobile money reduced the fixed cost of providing financial services to many people in Africa and boosted financial inclusion — traditional measures would have missed this leapfrogging.
- Leapfrogging observable in developing countries via mobile payments (Kenya), digital land registration (India), e-commerce (China).
- P2P lending provides loans faster, at lower cost, and without collateral requirements benefiting low-income households and SMEs; mobile point-of-sale and e-payments help SMEs document sales and obtain credit.

### Policy implications (Section VI) — recommendations and practices
- Main policy-relevant findings:
  - Fintech positively correlated with improving digital financial inclusion, and benefits poor and rural populations, but does not address the gender divide.
- Central bank interviews (Asia) indicated:
  - Digitalization efforts are often decentralized, complicating overarching strategies to reduce exclusion among vulnerable groups.
  - Two-track approach adopted: go digital for those willing and able; ensure those willing but unable retain access to cash and personalized help — aim to be cash-light but not completely cashless.
  - Examples from Singapore:
    - “Go Digital Campaign” funds SME digitalization.
    - “Hawkers Go Digital” provides cash incentives for hawkers to adopt digital payments.
    - “Seniors Go Digital Campaign” provides one-on-one guidance to seniors on using mobile phones for digital payments.
- Additional policy measures highlighted:
  - Raise financial and digital literacy.
  - Address infrastructure issues: broadband, wifi availability, last-mile challenges.
  - Complement Fintech development with targeted policies to improve women’s access to internet and to address discrimination, social norms, or laws that disadvantage women.
  - Address algorithmic bias and prediction precision problems, potentially moving fair lending laws toward outcome-focused, welfare-oriented approaches.
  - Recognize not all gaps should or can be closed; consider voluntary exclusion and structural reasons, and weigh risks and costs of closing gaps.
- Further research recommended to understand determinants of the Fintech gender gap (preferences, risk aversion, prior negative experiences, social norms, legal constraints).

### Conclusions (Section VII)
- Three key conclusions:
  - Fintech has a positive correlation with financial inclusion, stronger when using digital financial inclusion measures versus traditional measures.
  - Fintech has played a positive role in bridging the digital access gap between rural and rich-poor populations.
  - Fintech has no correlation with the gender digital gap.
- Policy takeaway:
  - Leave no one behind when promoting Fintech services; complement Fintech with targeted initiatives to close the gender gap in access to financial services.

*Source: Executive Summary, Sections V–VII, Appendices A–D — wpiea2022080-print-pdf (IMF Working Paper).*

### Executive Summary ......................................................................................................

### Executive Summary

### I. Introduction
- Section title present in source: "I.Introduction"
- Page number indicated in source: 5

### II. RECENT TRENDS AND LITERATURE
- Section title present in source: "II.RECENT TRENDS AND LITERATURE"
- Page number indicated in source: 5

### III. Mind The Gap
- Section title present in source: "III.Mind The Gap"
- Page number indicated in source: 10

### IV. Data and Empirical Approach
- Section title present in source: "IV.Data and Empirical Approach"
- Page number indicated in source: 13

### V. Empirical Assessment
- Section title present in source: "V.Empirical Assessment"
- Page number indicated in source: 17
- Subsections listed in source:
  - A.Does Fintech Promote Financial Inclusion? (page 17)
  - B.Does Fintech Bridge the Gender, Class and Rural Divides? (page 19)

### VI. Policy Implications
- Section title present in source: "VI.Policy Implications"
- Page number indicated in source: 22

### VII. Conclusion
- Section title present in source: "VII.Conclusion"
- Page number indicated in source: 23

### Appendices and References
- Appendix A. Data Definitions and Source (page 24)
- Appendix B: Methodology Underlying the Compilation of the Indices (page 26)
- References (page 30)

### Glossary (terms listed in source)
- ASEAN Association of Southeast Asian Nations
- DAI Digital Adoption Index
- DII Digital Intelligence Index
- EAP East Asia and Pacific
- IMF International Monetary Fund
- LAC Latin America and Caribbean
- MEA Middle East and North Africa
- OECD The Organization for Economic Co-operation and Development
- SSA Sub-Saharan Africa
- WB The World Bank

*Source: Executive Summary, wpiea2022080-print-pdf (IMF Working Paper) — table of contents and glossary as provided in the source PDF.*

### Executive Summary

### wpiea2022080-print-pdf - Executive Summary

### Executive summary: core questions and conclusions
- Research questions:
  - Does Fintech improve digital financial inclusion?
  - Are there segments of society not included because they do not have the capacity and means to adopt Fintech (women, the poor, people in rural areas)?
- Main empirical conclusions (based on panel regression):
  - Fintech has a positive correlation with financial inclusion, with the largest correlation occurring when digital financial inclusion is used.
  - Fintech has a negative association with the class divide and rich-poor divide (i.e., it has the potential to make it better).
  - Fintech has no impact on the gender digital gap.
- Policy implication highlighted:
  - Fintech alone may not be sufficient to close the gender gap in access to financial services; Fintech development may need to be complemented with targeted policy initiatives aimed at addressing the gender gap directly, and at changing attitudes and social norms across demographic groups.

### Introduction: framing and measurement challenges
- Observations:
  - Fintech is transforming financial services by leveraging technology and cloud-based data to offer products better tailored to consumers’ needs at lower cost.
  - Few studies provide direct empirical evidence of Fintech’s impact on financial inclusion due to challenges measuring Fintech and the “digital” aspect of financial inclusion, patchy data, and short time series.
- Measurement approach used in this paper:
  - Use of new digital financial inclusion indicators and various Fintech proxies, specifically: Digital Financial Inclusion Index (Sahay et al. 2020), Fletcher School’s Digital Intelligence Index (DII), and World Bank’s Digital Adoption Index (DAI).

### Recent trends, literature, and the “dark side” of Fintech
- Scope of Fintech (categories identified):
  - (i) credit, deposits, and capital-raising services;
  - (ii) payments, clearing and settlement services, including digital currencies;
  - (iii) investment management services;
  - (iv) insurance.
- Empirical and regional patterns:
  - Significant increase in mobile accounts and mobile money usage between 2014 and 2017.
  - Digital financial inclusion increased between 2014 and 2017 even where traditional inclusion stalled or declined.
  - In eight countries (including Zimbabwe, South Africa, Nigeria) improvements in financial inclusion were entirely driven by Fintech (Sahay et al. 2020).
- Divergent study findings on gender effects:
  - Sahay et al. (2020) found digital financial inclusion gender gaps are lower on average than traditional financial inclusion gender gaps overall and in Africa, the Middle East and Central Asia; but higher in Asia-Pacific and LAC.
  - Chen et al. (2021) found a persistent “Fintech gender gap”; gender gap smaller for products that complement traditional banking.
  - BIS working paper reported 29% of men use Fintech products and services versus 21% of women; Fintech gender gap 8 percent point.
- The “dark side” of Fintech includes:
  - Exclusion of women, the aged, the poor, and minority groups;
  - Algorithm biases and predatory lending practices;
  - Risk that digital delivery excludes those without access to digital infrastructure, financial/digital literacy, or trust.

### Mind the gap: measured divides and regional magnitudes
- Gender gap:
  - The gender gap exists across all regions and across measures (accounts, used internet to pay bills, made or received digital payments).
  - Average gender gap across all regions, using the accounts measure, is 11 percent points.
  - Example: in MEA the gender gap as measured by number of accounts is 19 percent points.
- Rich-poor (class) gap:
  - The average rich-poor gap is 15 percent points across all regions.
  - Regional magnitudes using accounts measure: LAC 21 percent points, EAP 19 percent points.
  - In SSA, MEA and SAS the rich-poor gap is smaller and narrows when digital measures are used.
- Other descriptive facts:
  - In the US 93% of adults have a bank account; there exists a 13 percent point gap between the wealthiest households and the poorest.
  - Internet access disparity cited: 48% of women have access to the internet compared to 55% for men globally (Alliance for affordable internet, 2021).

### Data, indices, and empirical approach
- Fintech proxies used (four):
  - Leapfrog (proxied by mobile phones/fixed line subscriptions)
  - Venture capital raised for Fintech companies (as % of GDP)
  - Total amount of funds raised through P2P lending platforms (as % of GDP)
  - Mobile money accounts
- Financial inclusion indicators used:
  - Digital Financial Inclusion Index (Sahay et al. 2020) — covers 52 EMDEs for 2014 and 2017; three-stage principal component analysis creating access and usage measures, and comprehensive financial inclusion index.
  - Fletcher School’s Digital Intelligence Index (DII) — covers 90 economies from 2008 to 2019 using 160 indicators across supply, demand, institutional environment, and innovation and change.
  - World Bank’s Digital Adoption Index (DAI) — composite index covering 171 countries for 2014–2016 using 16 indicators across businesses, people and governments.
- Measures of digital divide constructed:
  - Gender divide: deviation of female digital usage from male (higher is better = less divide).
  - Rural divide: deviation of rural digital usage from the average (higher is better).
  - Class divide: deviation of the bottom 40% from the top 60%.
- Summary statistics for indices:
  - Traditional Financial Inclusion Index — Mean 0.33; Standard deviation 0.22; Min-Max 0.0-1.0
  - Digital Financial Inclusion Index — Mean 0.36; Standard deviation 0.20; Min-Max 0.0-1.0
  - Digital Intelligence Index — Mean 0.51; Standard deviation 0.19; Min-Max 0.1-1.0
  - WB Digital Adoption Index — Mean 0.5; Standard deviation 0.19; Min-Max 0.1-0.8
- Correlation matrix excerpts (as presented):
  - No. of accounts row: X 0.56 0.89 0.35 0.87 0.86
  - Branches row: X 0.67 -0.04 0.38 0.50
  - Traditional Financial Inclusion Index row: X 0.22 0.76 0.82
  - Digital Financial Inclusion Index row: X 0.33 0.42
  - Digital Intelligence Index vs World Bank Digital Adoption Index: X 0.88
- Interpretation of correlations:
  - Highest correlation is between the Digital Intelligence Index and the World Bank Digital Adoption Index.
  - Lowest correlation (0.2) is between the IMF traditional Financial Inclusion Index and the IMF-Fintech Driven Index.
  - The IMF-Fintech Driven Index has only a 0.3 and 0.4 correlation with the Digital Intelligence Index and the Digital Adoption Index respectively.
  - Correlations indicate Fintech may substitute for bank branches (Fintech financial inclusion indicator negatively correlated with branches).

### Findings on substitution, complementarity, and exclusion
- Fintech vs traditional infrastructure:
  - Fintech can have negative correlations with traditional measures of access and usage (ATMs, bank branches, bank accounts per capita), suggesting substitution effects in some contexts.
  - Evidence that in some settings Fintech complements traditional banking (e.g., Solomon Islands remittances).
- Differential effects across regions and products:
  - Borrowing is much less widespread than account holding across regions; Fintech innovations (mobile money, P2P lending) have filled some lending gaps.
- Exclusion dimensions emphasized for policy focus:
  - Gender: persistent Fintech gender gap driven by attitudes, preferences, security concerns, lower willingness to adopt new financial technologies, and access constraints.
  - Aging: ten factors contributing to elderly financial exclusion include low digital capability, low financial literacy, cognitive decline, physical decline, social isolation, etc.
  - Algorithmic biases: fairness and accuracy issues in data and algorithmic credit scoring can disadvantage lower income and minority groups as well as women; lack of data for minority and low-income groups reduces prediction precision.

### Policy recommendations and implications (as stated)
- Fintech expansion should be paired with targeted policy measures to address exclusionary gaps, particularly:
  - Targeted initiatives aimed at addressing the gender gap directly.
  - Policies to change attitudes and social norms across demographic groups.
  - Attention to digital infrastructure, financial and digital literacy, trust, and inclusive product design to avoid widening divides.
- Suggested broader regulatory/policy concerns:
  - Address algorithmic bias and prediction precision problems, potentially moving fair lending laws toward outcome-focused, welfare-oriented approaches.
  - Recognize not all gaps should or can be closed; consider voluntary exclusion and structural reasons, and weigh risks and costs of closing gaps.

*Source: IMF Working Paper — Executive Summary (wpiea2022080-print-pdf).*

### Section V), primarily through the use of mobile money which has reduced the fixed cost of providing financial

### wpiea2022080-print-pdf - Section V), primarily through the use of mobile money which has reduced the fixed cost of providing financial

### Empirical approach
- Model specification (fixed-effect panel with robust standard errors):
  - FI_it = c + β1 X_it + β2 Fintech_it + α_i + ε_it  (equation (1))
  - XDIVIDE_it = β1 + β2 X_it + β3 Fintech_it + α_i + ε_it  (equation (2))
- FI is either digital or traditional financial inclusion; controls (Xs) include per capita income, credit to private sector, education level, regulatory quality, population, internet penetration, and mobile phone penetration.
- Three digital divides (Fletcher School): gender divide, class divide, rural divide. Divide variables reversed in regressions so that a lower number (i.e. a smaller gap). means an improvement.
- Fintech proxies: Leapfrog (ratio of mobile phones to fixed lines), Capital Invested (venture capital raised for Fintech), Capital raised (P2P lending as percent of GDP), Business Volume (digital lending/capital raising).

### Does Fintech promote financial inclusion? — Key empirical results (summary of Table 3)
- Fintech proxied by leapfrog and capital raised:
  - Positive and statistically significant in most regressions with digital financial inclusion and the Digital Intelligence Index (DII).
  - Leapfrogging has no significant effect with traditional financial inclusion (bank branches); capital raised has no correlation with the traditional financial inclusion indicator (Eq 2, 5, 8).
  - Leapfrogging shows negative correlation with traditional indicators and branches (implying Fintech is a substitute for bank branches).
- Specific reported coefficients (Table 3):
  - Population: 2.551*** (Digital FI, col 1); -5.835 (Branch, col 2)
  - Regulatory Quality: -0.077 (col 1); 0.056*** (DII, col 4)
  - Credit/GDP: -0.384 (col 1); 8.262*** (Branch, col 2)
  - Education (secondary): 0.011* (col 1); -0.112*** (Branch, col 2)
  - Cellphone/population: 0.003*** (col 1)
  - Internet/population: 0.074* (col 1); 0.149*** (Digital FI, col 3)
  - Leapfrogging (cell/fixed line): 0.037*** (Digital FI, col 1); -0.172** (Branch, col 2)
  - Fintech (Capital Invested): 0.013* (Digital FI, col 3); 0.004*** (DII, col 4)
  - Fintech (Capital raised, P2P): -0.014 (Digital FI, col 6); 0.004*** (DII, col 7); -0.206** (Branch, col 8)
- Model fit and sample sizes (selected):
  - Observations: 64 (col 1), 950 (col 2), 33 (col 3), 369 (col 4)
  - R-squared: 0.786 (col 1), 0.197 (col 2), 0.910 (col 3), 0.692 (col 4)
  - Number of ctry: 36 (col 1), 152 (col 2), 19 (col 3), 65 (col 4)
- Interpretation:
  - Leapfrog variable has the highest positive effect on digital financial inclusion (Digital Inclusion Index) and negative association with traditional branches — evidence of technological leapfrogging substituting for branch infrastructure.
  - Capital invested significant at 10% for digital financial inclusion (Eq 3).
  - P2P lending (capital raised as % of GDP) negatively affects branches, positively correlates with DII (Eq 7), but no significant relationship with the Digital Financial Inclusion Index (Eq 6) — implying complementarity with traditional bank lending but substitution for branches, and helping fill the lending gap for low-income households and SMEs.

### Fintech and the rural, class, and gender divides — Key results
- Estimation framework controls for trust index, regulatory quality, education; gender divide regressions also include gender inequality index.
- Main findings:
  - Fintech proxies are significant and negatively correlated with the rural divide and the class divide — implying Fintech can improve rural and class divides.
  - Fintech proxies show no significant correlation with the gender divide.
- Selected coefficient highlights (Tables 4–6):
  - Rural Divide (Table 4):
    - Trust index: -0.382*** (col 1)
    - Leapfrog: -3.039* (col 2)
    - Digital financial inclusion: -4.948** (col 3)
    - Digital adoption index: -7.138** (col 4)
    - Fintech (Capital invested): -0.094** (col 5)
    - Fintech (Capital raised, P2P): -0.069** (col 5)
    - Observations: 147 (col 1); R-squared: 0.286 (col 1)
  - Class Divide (Table 5):
    - Trust index: -0.575*** (col 1)
    - Leapfrog: -2.357*** (col 2)
    - Digital financial inclusion: -3.857*** (col 3)
    - Digital adoption index: -6.282*** (col 4)
    - Fintech (Capital invested): -0.056*** (col 5)
    - Fintech (Capital raised, P2P): -0.060*** (col 5)
    - Observations: 147 (col 1); R-squared: 0.450 (col 1)
  - Gender Digital Divide (Table 6):
    - Regulatory quality: -0.416* (col 1); -0.443** (col 2); -0.416** (col 3)
    - Gender inequality: 4.555 (col 1); 3.553 (col 2); 1.561 (col 3)
    - Digital adoption index: -0.882 (col 2)
    - Fintech (Capital invested): 0.013 (col 2)
    - Fintech (Capital raised, P2P): -0.004 (col 3)
    - Observations: 41 (col 1); 137 (col 2); 98 (col 3); R-squared: 0.325 (col 1)
- Robustness:
  - Results using raw gap measures (e.g., digital payments female – digital payments male) yield similar conclusions (Appendix C).
- Caveat:
  - Explanatory powers of these variables are rather low; further analysis with more variables is suggested as new data become available.

### Mechanisms and examples
- Mobile money reduced the fixed cost of providing financial services to many people in Africa and boosted financial inclusion — traditional measures would have missed this leapfrogging.
- Leapfrogging observable in developing countries via mobile payments (Kenya), digital land registration (India), e-commerce (China).
- P2P lending provides loans faster, at lower cost, and without collateral requirements benefiting low-income households and SMEs; mobile point-of-sale and e-payments help SMEs document sales and obtain credit.

### Policy implications (Section VI)
- Main policy-relevant findings:
  - Fintech positively correlated with improving digital financial inclusion, and benefits poor and rural populations, but does not address the gender divide.
- Central bank interviews (Asia) indicated:
  - Digitalization efforts are often decentralized, complicating overarching strategies to reduce exclusion among vulnerable groups.
  - Two-track approach adopted: go digital for those willing and able; ensure those willing but unable retain access to cash and personalized help — aim to be cash-light but not completely cashless.
  - Examples from Singapore:
    - “Go Digital Campaign” funds SME digitalization.
    - “Hawkers Go Digital” provides cash incentives for hawkers to adopt digital payments.
    - “Seniors Go Digital Campaign” provides one-on-one guidance to seniors on using mobile phones for digital payments.
- Additional policy measures highlighted:
  - Raise financial and digital literacy.
  - Address infrastructure issues: broadband, wifi availability, last-mile challenges.
  - Complement Fintech development with targeted policies to improve women’s access to internet and to address discrimination, social norms, or laws that disadvantage women.
  - Further research recommended to understand determinants of the Fintech gender gap (preferences, risk aversion, prior negative experiences, social norms, legal constraints).

### Conclusions (Section VII)
- Three key conclusions:
  - Fintech has a positive correlation with financial inclusion, stronger when using digital financial inclusion measures versus traditional measures.
  - Fintech has played a positive role in bridging the digital access gap between rural and rich-poor populations.
  - Fintech has no correlation with the gender digital gap.
- Policy takeaway:
  - Leave no one behind when promoting Fintech services; complement Fintech with targeted initiatives to close the gender gap in access to financial services.

*Source: IMF Working Paper (Section V–VII, Tables 3–6, Appendix A definitions).*

### Appendix B:  Methodology Underlying the

### Appendix B: Methodology Underlying the Compilation of the Indices

### Digital Intelligence Index (Digital Evolution)
- Purpose and scope:
  - Data-driven holistic evaluation of the progress of the digital economy across 90 economies.
  - Combines 160 different indicators into four key drivers: supply conditions, demand conditions, institutional environment, and innovation and change.
- Structure:
  - Four-level hierarchy: indicators → clusters → components → drivers.
  - 160 indicators aggregated into clusters that illuminate 35 aspects of digitalization, rolled into 13 higher-order components, feeding into the four drivers.
- Weighting and aggregation:
  - Central hypothesis: the four drivers are of equal importance; model accords equal weight to all four drivers.
  - Indicator, cluster, and component weights are determined according to three factors: data quality, data centrality, and the strength of data collection methods.
  - Indicators aggregated into clusters, components, drivers, and final scores; final scores rescaled to fit a 0 to 100 range.
- Momentum scoring:
  - Momentum scores generated by applying the compound annual growth rate formula (CAGR) on final index scores across the time window.
  - After calculating index growth rates for each economy, CAGRs are rescaled on a 0 to 100 scale.
  - Momentum scores are relative, like the final index scores.
- Quality assurance and validation:
  - Multiple statistical tools used: data cleaning, variance analysis, regression analysis, and simulations.
  - Index scores stress-tested at multiple levels to ensure validity and robustness.

### World Bank Digital Adoption Index (DAI)
- Scope and scale:
  - Described as a supply-side measure.
  - DAI is a worldwide index that measures countries’ digital adoption across three dimensions: people, government, and business.
  - Covers 180 countries on a 0–1 scale (text also states: DAI measures adoption in 171 countries in a later methodological description).
  - Two observations available for most countries: 2014 and 2016.
- Aggregation formula:
  - Overall DAI is the simple average of three sub-indexes:
    - DAI (Economy) = DAI (Businesses) + DAI (People) + DAI (Governments)
  - Each sub-index assigned equal weight.
- Sub-index composition:
  - DAI (Business): simple average of four normalized indicators:
    - percentage of businesses with websites
    - number of secure servers
    - download speed
    - 3G coverage in the country
  - DAI (People): simple average of two normalized Gallup World Poll indicators:
    - mobile access
    - internet access at home
  - DAI (Governments): simple average of three sub-indices:
    - core administrative systems
    - online public services (data from the UN’s Online Service Index)
    - digital identification
    - Sixteen indicators used to develop the three sub-indices of the Government cluster.
- Data and limitations:
  - Much of the data was collected as part of the WDR 2016 preparation; indicators need regular updating.
  - Additional indicators suggested: ministry-specific management information systems (education, health, welfare payments).
  - Business cluster’s four indicators are limited; firm-specific variables (enterprise resource planning, cloud computing, supply chain management software, e-purchases and e-sales) are largely available only for advanced economies and need expansion to the developing world.
  - Methodology is flexible to accommodate new digital technologies (e.g., mobile money, big data) and to drill down to more disaggregated levels (e.g., DAI for e-retail or digital ID).

### Appendix C: Robustness Check With Raw Gap Measure (Regression Evidence)
- Alternative dependent variables used:
  - Gaps from Findex Survey: account gap (“% of male with account - % of female with account”), gap in “use internet to pay bill”, and gap in “made/received digital payment”.
  - Class divide measured as “60% richest – 40% poorest”.
  - These gap indicators used as dependent variables in panel regressions.
- Summary statement of results:
  - The results in Table A1 and Table A2 confirm earlier findings about the gender digital gap and class divide.
  - In some cases, a wider gap is associated with Fintech (as reported in the tables).
- Table A1: Fintech and Gender Digital Divide — selected model diagnostics and counts
  - Observations by column: 22, 22, 22, 24, 24, 24, 51, 51, 51
  - Number of ctry by column: 11, 11, 11, 12, 12, 12, 27, 27, 27
  - R-squared by column: 0.328, 0.331, 0.130, 0.428, 0.460, 0.192, 0.117, 0.070, 0.102
  - Adj. R-squared by column: 0.169, 0.173, -0.0750, 0.307, 0.346, 0.0216, 0.0405, -0.0112, 0.0235
  - Notable coefficient significance markers in the table:
    - Trust index coefficients marked ** or *** in several specifications.
    - Digital Financial Inclusion: coefficient 0.149** in one column.
    - Mobile Money Account: coefficient 0.462*** in one column.
    - Fintech (Capital raised, P2P): coefficients reported (e.g., -0.000, -0.001, -0.006) with standard errors shown.
  - Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.
- Table A2: Fintech and Class Digital Divide — selected model diagnostics and counts
  - Observations by column: 22, 22, 22, 24, 24, 24, 51, 51, 51
  - Number of ctry by column: 11, 11, 11, 12, 12, 12, 27, 27, 27
  - R-squared by column: 0.304, 0.610, 0.460, 0.296, 0.279, 0.116, 0.217, 0.048, 0.128
  - Adj. R-squared by column: 0.140, 0.518, 0.333, 0.148, 0.127, -0.0701, 0.149, -0.0343, 0.0518
  - Notable coefficient significance markers in the table:
    - Regulatory quality coefficients show significance in several specifications (e.g., 0.117**, 0.065**, 0.088*).
    - Trust index coefficients include significant negative estimates in several specifications (e.g., -0.066**, -0.078***, -0.103***).
    - Digital Financial Inclusion: coefficient 0.393*** in one column.
    - Mobile Money Account: coefficient 1.051** in one column.
    - Fintech (Capital raised, P2P): coefficients reported (e.g., 0.001, 0.005, -0.007) with standard errors shown.
  - Robust standard errors in parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.
- Interpretation provided by authors:
  - The gap-based regressions corroborate earlier findings on gender and class digital divides; in some cases Fintech is associated with a wider gap.

### Appendix D: Interview Questions with Policymakers (selected questions)
- In the design of the digitalization/fintech national strategy, have you considered the segments of society that might be excluded e.g. the poor, aged, women, those in rural/remote areas?
- Who amongst the “excluded group” do you think are the most excluded and therefore needs most help?
- How do you take into account these gaps and what needs to be done to help those excluded or make sure that they can catch up?
- What are the consequences of ignoring this group?
- Would you consider slowing the pace of fintech development in your country as part of your strategy to “wait” for them to catch up?
- Would additional data initiatives help in the design of policies to mitigate financial exclusion? If so, any specific examples?

*Source: Appendix B–D, Methodology and supporting material from the IMF Working Paper (wpiea2022080-print-pdf).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022080-print-pdf.pdf_