Does Disproportionate Financial Inclusion Reduce Gender and Income-Group Inequality? Global Evidence

Abstract

This paper investigates whether countries’ investments in financial inclusion, beyond what their economic capacity would predict, help mitigate gender and income-group gaps in financial opportunities. We construct a novel index, the Abnormal Financial Inclusion Index (AbFII), as the residuals from a regression of financial inclusion on GDP per capita to isolate country-specific efforts in advancing financial inclusion. Using data from 100 countries between 2014 and 2021, we find that higher AbFII is associated with lower gender and rich-poor inequalities in financial inclusion, particularly in low-income and high-inequality countries. A one-standard-deviation increase in AbFII corresponds to a 1.2 percentage point reduction in gender and income-group disparities in financial access. We also find that AbFII predicts subsequent improvements in broader gender inequality, especially in reproductive health. The results are robust to various controls and remain consistent under instrumental variable analysis using broadband penetration as an instrument. Our findings suggest that financial inclusion policies exceeding the level expected by economic development can significantly reduce social disparities.

1. Introduction

Financial inclusion refers to the accessibility and availability of financial services to all segments of the population, particularly those who are unbanked or underserved. The World Bank Group (2022) defines it as “individuals and businesses having access to useful and affordable financial products and services that meet their needs delivered in a responsible and sustainable way.” Financial inclusion implies efforts by governments and institutions to reduce disparities in financial access across gender and income groups. It thus has important social implications, particularly for women, low-income individuals, and those in underdeveloped countries. The concept is multifaceted, encompassing a country’s overall level of inclusion and disparities across demographic groups. Achieving financial inclusion that is both broad and equitable requires attention to these group-based inequalities.

This paper empirically investigates whether a country’s investment in financial inclusion is associated with reduced disparities in access across gender and income groups. Reducing such disparities has attracted considerable attention from researchers and policymakers (Jain-Chandra et al., 2017; Fouejieu et al., 2020; Demirgüç-Kunt et al., 2013, 2015, 2022). The World Bank has identified universal access to transaction services as a milestone toward full inclusion. Similarly, one of which aims to reduce inequality among different groups across various dimensions (https://sdgs.un.org/goals, accessed on 13 May 2023).1

The main channel through which financial inclusion may reduce inequality is by easing barriers commonly faced by disadvantaged groups, such as physical distance from financial institutions and lack of formal documentation (Sioson & Kim, 2019; Hanmer & Elefante, 2019). Consequently, when developments in fintech make access to finance easier, the benefits are more likely to be brought to those most disadvantaged, thereby improving inequalities in financial inclusion. Several studies have reported that state-led improvements in financial inclusion have played a role in reducing the gender gap in financial inclusion, as seen in the case studies of developing countries (Mbiti & Weil, 2018).

Despite some anecdotal evidence and a few case studies on this issue, there is a lack of comprehensive cross-country empirical research on the relationship between the level of financial inclusion and inequality. In this paper, we aim to fill the gap. Using the World Bank database, we examine the impact of the national investment in financial inclusion on the global reduction in disparities in six financial inclusion measures across genders and dichotomous income groups within countries.

A key challenge in financial inclusion research is that a country’s level of financial inclusion is strongly correlated with its economic strength, making it difficult to isolate the pure effect of financial inclusion in empirical analysis.2 To address this, we propose a novel measure: the Abnormal Financial Inclusion Index (AbFII), defined as the residual from a regression of financial inclusion on GDP per capita.3 AbFII captures the portion of financial inclusion beyond what is predicted by income level, serving as a proxy for a country’s unexpected or policy-driven investment. For example, despite its modest income level, Thailand scores positively on AbFII due to initiatives such as expanding state-run digital banking and targeted financial access programs.

To summarize, this paper addresses two research objectives: (1) to develop a composite index that captures a country’s level of financial inclusion beyond what can be explained by its level of economic development and (2) to examine whether the index can explain the cross-country differences in gender and rich-poor inequalities scattered over the various financial inclusion metrics.

Using data from 100 countries from 2014 to 2021, we first construct the AbFII, which serves as a proxy for a country’s unexpected investment in financial inclusion beyond what its level of economic development would suggest. We find that countries with positively abnormal financial inclusion experience significantly lower gender and rich-poor inequalities in access to financial services. In other words, a disproportionately high financial inclusion level significantly reduces gender and income inequality. Our results remain robust to a wide range of country-level control variables as well as time and region fixed effects. Furthermore, while we document that gender and rich-poor inequality in financial inclusion is more severe in low-income countries and countries with high levels of overall gender and income inequality, we also find that the positive effect of AbFII on improving gender equality in financial inclusion is more pronounced in countries with a poorer level of overall gender inequality measure. This result implies that countries with more severe inequality may have incentives to make disproportionate investments in financial inclusion to address structural gaps.

In additional analyses, we explore whether AbFII is associated with improvements in overall gender inequality beyond the domain of financial inclusion. Using the UNDP’s Gender Inequality Index and its subcomponents, we find that higher AbFII values are significantly related to lower gender inequality in subsequent years, particularly in female reproductive health outcomes. These findings suggest that broader social outcomes may follow from improved financial access among disadvantaged groups. We also address potential endogeneity concerns through a two-stage least squares (2SLS) approach using broadband internet penetration as an instrument for AbFII. The IV results remain consistent with our baseline estimates, lending further support to the robustness of our findings.

Our main contribution lies in developing and applying AbFII, which isolates financial inclusion from general economic development. Unlike more complex machine learning or non-parametric approaches, AbFII offers a transparent and easily interpretable measure well-suited for cross-country empirical analysis and policy application. Additionally, while prior studies have primarily focused on financial inclusion’s overall benefits or country-specific case studies, our cross-country analysis provides the first systematic evidence on whether excess national investments in financial inclusion help reduce group-based disparities. Furthermore, we show that gender and rich-poor inequalities in financial inclusion remain significantly more severe in low-income countries. Lastly, our results suggest that the AbFII can explain meaningful variation in these inequalities across countries and point to the need for more targeted policies if the UN’s Sustainable Development Goals 5 and 10 are to be realized, particularly in developing and less developed economies.

The remainder of this paper is organized as follows. Section 2 introduces the background, presents global trends in selected financial inclusion metrics, and highlights the disparities across genders and income groups. In Section 3, we examine previous studies on the role of financial inclusion in economic development and poverty reduction, the dynamics of intra-country inequalities, and the evolution of financial inclusion measurement, emphasizing our study’s unique contribution. Section 4 and Section 5 describe our sample selection and empirical strategy, respectively. Section 6 details our findings and discusses their implications, and finally, Section 7 summarizes our contributions, reflects on policy implications, and suggests avenues for future research.

2. The Background

2.1. Global Financial Inclusion and Gender and Rich-Poor Inequalities

The World Bank Global Findex Database (https://www.worldbank.org/en/publication/globalfindex, accessed on 13 May 2023) provides global survey data on financial inclusion, covering indicators such as digital payments, account ownership, financial institution account ownership, debit or credit card ownership, saving, and borrowing. For the details of each measurement category, see Appendix A. In addition, we draw on UNDP data to contextualize recent trends in global financial inclusion and its association with two key dimensions of inequality: gender inequality and rich-poor inequality. We focus on the six financial inclusion measures for which data are available for most countries. Following established practice in the literature, we group these indicators into three dimensions for concise representation: (1) “access” to financial services (account ownership, financial institution account ownership, debit, or credit card ownership), (2) “usage” of financial services (saving, borrowing), and (3) “technology-based” financial services (digital payments). We collect data from the World Bank Global Findex Database on the six measures for 100 countries in 2014, 2017, and 2021, resulting in 300 country-year observations. Note that these three years are selected because the Global Findex data are released by the World Bank at three- to four-year intervals.

All six financial inclusion measures show a consistently increasing global trend across the three observed years (see Figure 1). This trend suggests that many countries worldwide have actively invested in improving financial inclusion over the past decade. However, progress toward achieving gender and income-group equality in financial inclusion remains unfulfilled. Gender inequality is measured as the percentage point difference between males and females in each country’s financial inclusion indicators. Similarly, we measure rich-poor inequality as the percentage point difference between individuals in the top 60 percent of the income distribution (rich) and those in the bottom 40 percent (poor) within each country. Both gender and income-group inequalities improved between 2014 and 2017 but worsened by 2021, indicating a reversal in recent years. While global efforts to promote gender equality led to positive outcomes in most financial inclusion dimensions from 2014 to 2017, except for debit or credit card ownership, these gains were largely reversed between 2017 and 2021, possibly due to the socio-economic impact of COVID-19 (see Figure 2). Rich-poor inequality tends to be more pronounced than gender inequality. Rich-poor disparities widened throughout the period for debit or credit card ownership and saving. In contrast, the other four financial inclusion indicators showed improvement from 2014 to 2017 but deteriorated from 2017 to 2021 (see Figure 3).

In summary, while the overall trend in financial inclusion is encouraging, the patterns in inequality reveal persistent and, in some cases, worsening gaps between men and women and between income groups despite increasing levels of access. The post-2017 reversal may suggest that, as many countries scaled up their financial inclusion initiatives, policy efforts were directed more toward expanding overall access rather than addressing group-specific disparities. This pattern implies that while financial inclusion increased in aggregate terms, inequality in access across gender and income groups may have been overlooked. It is noteworthy that, among all regions, Asian countries most clearly exhibited a pattern in which gender disparities worsened even as overall financial inclusion increased after 2017.

2.2. Financial Inclusion and Overall Gender Inequality

Financial inclusion is expected to improve not only gender inequality in financial access but also overall gender inequality. An improvement in gender inequality in financial inclusion means the disadvantaged groups have better access to finance, which may influence various gender gap metrics across society. As gender inequality spans multiple dimensions, it is difficult to define it. Therefore, this paper discusses overall gender inequality based on the Gender Inequality Index (GII), constructed and published by the United Nations Development Programme (UNDP).

According to UNDP’s Human Development Reports, GII reflects gender inequality in three major dimensions: reproductive health, empowerment, and the labor market. Female reproductive health is assessed using the maternal mortality ratio and adolescent birth rate, and the portion of the female population measures female empowerment with higher education and female shares of parliamentary seats. Lastly, the female labor force participation rate serves as the labor market indicator.

Previous studies suggest that reducing financial barriers is vital in improving the three dimensions of GII. Financial barriers hinder timely access to maternal health services (Comfort et al., 2013) and access to safe abortion (Saraswati, 2022), which are crucial to the prevention of pregnancy-related deaths in developing countries. Regarding adolescent pregnancy, Chandra-Mouli et al. (2013) and Orimaye et al. (2021) suggest that access to primary care physicians is critical for effective contraceptive counseling and follow-up procedures. The other two dimensions of GII, female empowerment, and labor market participation, are closely related to education opportunities, where financial barriers could also affect access to it. Numerous previous studies, such as Checchi et al. (2014), have found that an individual’s financial constraints significantly limit access to secondary education.

3. Literature Review

Research on financial inclusion spans various topics, including its potential to foster economic development, reduce poverty, and promote social equity. In this section, we group the existing literature into three thematic areas: (i) the role of financial inclusion in economic development, (ii) financial inclusion and demographic inequality, and (iii) methods for constructing financial inclusion indices.

3.1. Financial Inclusion and Economic Development

Early studies emphasized the link between financial inclusion and macroeconomic development. Increased access to financial services enables individuals and small businesses to manage resources more effectively, invest in income-generating activities, and improve their standard of living (The World Bank Group, 2014, 2022; Jain-Chandra et al., 2017; Qamruzzaman & Jianguo, 2017, 2018; Fouejieu et al., 2020; Chen & Yuan, 2021). In particular, financial inclusion has been associated with higher levels of savings and domestic capital accumulation, which are critical for long-term growth.

Digital financial technologies, particularly mobile banking, have expanded financial access in underserved regions (Beck et al., 2004; Cull et al., 2014; Preziuso et al., 2023). These technologies reduce transaction costs and geographic barriers, bringing basic financial services within reach of remote and low-income populations.

Some studies also note that the effectiveness of financial inclusion efforts often depends on the presence of supportive institutions and infrastructure (Cull et al., 2014; Preziuso et al., 2023).

3.2. Financial Inclusion and Demographic Inequality

Despite aggregate progress in financial inclusion, disparities across demographic groups remain persistent. Numerous studies document significant gender and income-group gaps in access to financial services (Demirgüç-Kunt et al., 2013, 2015, 2022; Aterido et al., 2013; Jain-Chandra et al., 2017; Ghosh & Vinod, 2017; Kaur & Kapuria, 2020). For example, Demirgüç-Kunt et al. (2013) find that gender inequality in account ownership is more severe in developing countries and that such inequality persists even after controlling for macroeconomic conditions. Moreover, they highlight how specific dimensions of financial behavior—such as saving and borrowing—can display different patterns of gender disparity across income levels.

Mbiti and Weil (2018), focusing on Kenya, show how the rapid spread of mobile banking improved women’s access to financial services and had downstream effects on household bargaining power and savings behavior. While promising, their findings suggest that technology-driven inclusion may need to be accompanied by complementary policy measures to ensure equitable access across social groups.

Several studies argue that gender and income inequality interact with structural barriers such as labor market segmentation, asset inequality, and legal restrictions (Bosma et al., 2021; Aslan et al., 2017; Hanmer & Elefante, 2019). For instance, Cicchiello et al. (2021) report that women, rural populations, and low-income individuals in developing countries remain disproportionately excluded from formal financial products, reinforcing broader patterns of exclusion and disadvantage.

3.3. Constructing Financial Inclusion Indices

Scholars have proposed various approaches to measuring financial inclusion, aiming to capture its multi-dimensional nature. One of the earliest efforts is Sarma (2008), who construct a composite index by averaging three normalized dimensions: banking penetration (e.g., number of bank branches per capita), availability (e.g., number of ATMs), and usage (e.g., deposit and credit accounts). While intuitive and easy to implement, this method’s equal-weighting scheme and sensitivity to variable selection raise concerns about robustness.

Later studies adopted statistical techniques such as principal component analysis (PCA) and exploratory factor analysis (EFA) to address these limitations. Le et al. (2019) use PCA to identify the dominant variance among financial inclusion indicators and reduce dimensionality. Chinnakum (2023) applies EFA to uncover latent factors linked to inclusion and poverty reduction. These approaches allow flexible weighting but require judgment in variable selection and interpretation.

Camara and Tuesta (2014) integrate demographic and socio-economic variables into a PCA-based index, finding high correlations with GDP per capita, education, and financial system size. Park and Mercado (2018) also use PCA and show that their index predicts growth and poverty reduction in middle- and upper-income countries but not low-income economies.

Despite these advances, most indices fail to separate financial inclusion from underlying economic development fully. Their construction methods often embed structural characteristics—such as income, education, and urbanization—into the inclusion index. As a result, empirical analyses using these indices may inadvertently conflate inclusion with macro-level development. Our study addresses this limitation by introducing an abnormal index that explicitly controls for economic strength, thereby isolating the component of financial inclusion that is not explained by GDP per capita. The differences between the indices used in prior studies and our financial inclusion index are summarized in

Table 1. Comparison of financial inclusion indices between prior studies and this study.

Study	Method	GDP Controlled	Key Focus
Sarma (2008)	Averaging	No	Overall financial access
Le et al. (2019)	PCA	No	Financial efficiency and stability
Chinnakum (2023)	EFA	No	Poverty and inequality reduction
Camara and Tuesta (2014)	PCA	No	Multi-dimensional access and demographics
Park and Mercado (2018)	PCA	No	Growth and poverty reduction
This Study (AbFII)	PCA & Regression residuals	Yes	Disproportionate national inclusion effort on inequality reduction

.

4. The Theoretical Framework

4.1. The Construction of the Financial Inclusion Index (FII)

Given the multi-dimensional nature of financial inclusion, deriving a single composite financial inclusion index is a critical component in our empirical strategy. As discussed in the literature review, the existing literature has primarily taken two approaches to constructing financial inclusion indices: (1) constructing an index computing the average of selected metrics (e.g., Sarma, 2008) and (2) extracting the principal components of the metrics via an exploratory factor analysis (EFA) method or a principal component analysis (PCA) method (e.g., Le et al., 2019; Chinnakum, 2023). While the first approach is simple and intuitive, its validity depends on the soundness of the selected variables in the index. Due to this concern, our framework adopts the second approach to construct a financial inclusion index. In our analysis, we choose EFA over PCA to identify latent constructs that drive comovements among financial inclusion indicators rather than merely reduce dimensionality.

We conduct an exploratory factor analysis using the six financial inclusion variables to estimate our financial inclusion index (FII).

Table 2. Exploratory factor analysis.

Panel A: Factor identification
Factor	Eigenvalue	Difference	Proportion	Cumulative
1	215.699	212.993	0.980	0.980
2	2.706	0.518	0.012	0.993
3	2.188	1.719	0.010	1.002
4	0.469	0.780	0.002	1.005
5	−0.312	0.384	−0.001	1.003
6	−0.695		−0.003	1.000
Panel B: Factor loadings
Variable	Communality	Weight	Adj. Weight
DP	0.939	16.537	0.158
ACCT	0.971	34.656	0.330
FIN_ACCT	0.968	30.993	0.295
CARD	0.938	16.094	0.153
SAVING	0.725	3.635	0.035
BORROW	0.671	3.036	0.029

presents the result of the EFA, where Panel A reports the factors identified in the analysis, and Panel B shows the factor loadings of the first factor. The results in Panel A indicate that the first factor alone explains 98% of the total variance, suggesting it is a strong composite index for financial inclusion. In Panel B, the communality, the proportion of variance in one variable accounted for by an underlying element common to all variables in a set, is very high across all six variables, confirming that the first factor adequately captures the underlying structure shared among them.

One of the key challenges in financial inclusion research is the strong correlation between a country’s financial inclusion level and its overall economic development. This correlation makes it difficult to determine whether observed effects are attributable to financial inclusion or confounded by underlying economic conditions. To address this issue, we regress the financial inclusion index (FII) on GDP per capita, a widely used proxy for a country’s economic development, by year and use the residuals from this regression as our primary explanatory variable. We refer to this residual component as the Abnormal Financial Inclusion Index (AbFII).

$$AbFI{I}_{c,t}=FI{I}_{c,t}-(\widehat{{\beta }_0}+\widehat{{\beta }_1}\cdot GD{P}_{c,t})$$

(1)

β₀ and β₁ are estimated coefficients from the year-specific regression analysis of FII on GDP, and AbFII_c,t denotes the abnormal financial inclusion index of country c in year t. As expected, the adjusted R-squared values from the regressions (not reported in tables) are high, 0.796 for 2014 and 0.787 for 2017, indicating that GDP per capita explains most of the variation in FII. AbFII captures the portion of financial inclusion that exceeds what would be expected given a country’s economic development level. In other words, we interpret AbFII as a proxy for a country’s disproportionate or unexpected investment, effort, or policy emphasis toward expanding financial inclusion beyond what its income level alone would predict.4 This interpretation is particularly relevant in contexts where governments, regulators, or financial sectors implement targeted programs, subsidies, or infrastructure improvements to boost access to financial services beyond natural market-driven growth. In this paper, AbFII allows us to examine cross-country heterogeneity in gender and income-group disparities in financial inclusion independent of economic strength. Notably, GDP per capita can still be included as a control variable in our main regressions, as AbFII is orthogonal to it by construction.

4.2. Empirical Specification

We employ the following regression model specification to identify the causal effect of AbFII on gender and rich-poor inequalities in financial inclusion:

$$\begin{array}{ll}{Inequality}_{c,t+3}& ={\beta }_0+{\beta }_1{AbFII}_{c,t}+{\beta }_2\log {\left(GDP\right)}_{c,t}+{\beta }_3{GROWTH}_{c,t}\\ & +{\beta }_4\log (POP{)}_{c,t}+{\beta }_5{GE}_{c,t}+{\beta }_6{ROL}_{c,t}+{\sum }_t{Year}_t\\ & +{\sum }_r{Region}_r+{ϵ}_{c,t}\end{array}$$

(2)

The dependent variable, Inequality_c,t+3 denotes either the gender or the rich-poor gap in a financial inclusion metric of country c in year t + 3. Since we consider both gender and income-group disparities across six financial inclusion metrics, a total of 12 dependent variables are examined in this study. All explanatory variables are lagged by three years to capture better the causal effects of abnormal financial inclusion, which will likely emerge gradually over time. We acknowledge, however, that using lagged regressors alone may not fully resolve endogeneity concerns. Therefore, we perform a two-stage least squares (2SLS) estimation in the robustness analysis section to further address potential endogeneity issues.

The coefficient of primary interest is β₁, which captures the effect of AbFII_c,t, the abnormal financial inclusion index of country c in year t. As described earlier, AbFII represents the residual component of financial inclusion that is not explained by economic development. Following standard practice in the literature, we include a vector of country-level control variables. They control the country’s economic strength, size, and governance and include the log of GDP per capita (log(GDP)_c,t), the GDP growth rate (GROWTH_c,t), the log of population in millions (log(POP)_c,t), the government effectiveness index (GE_c,t), and the rule of law index (ROL_c,t).5 To account for unobserved heterogeneity, we include year- and region-fixed effects, where countries are grouped into seven global regions.6 Combining the fixed effects of year and region ensures that our estimates are robust to the unobservable and omitted effect of year and region. Lastly, while our main results are not based on bootstrapped standard errors, the use of bootstrapped standard errors does not materially affect the statistical significance of our findings.

5. Sample and Data Sources

We collect financial inclusion data from the World Bank Databank (https://databank.worldbank.org/, accessed on 24 May 2023). World Bank began publishing the Global Findex series in 2014 and has subsequently updated it twice, in 2017 and 2021. The Findex series covers 160 countries. The dataset covers approximately 160 countries. For our analysis, we construct a balanced panel of 100 countries with complete data availability across all three years. For instance, we exclude the mobile money account indicator, despite its usefulness as a proxy for fintech adoption, due to a high proportion of missing values (approximately 45%). A notable strength of the Global Findex Database is that it provides disaggregated data by subgroups such as gender (female, male), income (poorest 40%, richest 60%), location (urban, rural), and age categories. All subseries used in our analysis contain no missing values for the gender and income subgroups, enabling a consistent global investigation of financial inclusion disparities across these key dimensions.

Our final sample consists of 300 country-year observations based on a balanced panel of 100 countries from 2014, 2017, and 2021. The six financial inclusion indicators that meet our sampling criteria are: (1) the percentage of the population using a digital payment service (DP); (2) the percentage of the population having a bank account or using a mobile money service (ACCT); (3) the percentage of the population having a financial institution account (FIN_ACCT); (4) the percentage of the population having a debit or credit card (CARD); (5) the percentage of the population saving money at a financial institution (SAVING); and (6) the percentage of the population borrowing money from a financial institution (BORROW). Although the six indicators are highly correlated and exhibit high communalities, we categorize them into three conceptual dimensions. DP is considered a proxy for technology-based financial inclusion. ACCT, FIN_ACCT, and CARD are proxies for access to financial services. SAVING and BORROW represent the usage of financial services. We use these six indicators to construct our financial inclusion index (FII). In contrast, we use their subgroup variants (female, male, poorest 40%, and richest 60%) to measure gender and income-group disparities in financial inclusion.

We also collect country-level control variables from multiple sources. The country’s GDP per capita (GDP), GDP growth rate (GROWTH), population (POP), and the Gini coefficient (GINI) data are obtained from the World Bank database. Worldwide Governance Indicators (WGI) variables such as the country’s Rule of Law Index (ROL) and Government Effectiveness Index (GE) are also obtained from the World Bank database. According to the WGI documentation, the ROL measures the extent to which citizens have confidence in and comply with the rules of society. At the same time, GE reflects perceptions of the quality of public services, the competence and independence of the civil service, and the effectiveness of government policy implementation. Lastly, we obtain data on the Gender Inequality Index (GII) from the United Nations Development Programme (UNDP).

6. Results

6.1. Descriptive Statistics

Table 3. Descriptive statistics.

		Variables	Mean	Std. Dev.	Median	Min.	Max.
Financial Inclusiont		FII	0.595	0.271	0.614	0.079	0.979
		AbFII	0.000	0.131	0.010	−0.314	0.375
Inequalityt+3	G-gap	DP (%p)	5.64	7.06	4.25	−6.83	29.37
		ACCT (%p)	5.21	7.06	3.36	−6.98	29.18
		FIN_ACCT (%p)	5.09	6.91	3.17	−6.53	29.01
		CARD (%p)	5.66	6.78	4.33	−6.06	28.32
		SAVING (%p)	4.06	4.22	4.45	−7.76	15.72
		BORROW (%p)	4.20	5.10	3.68	−7.52	22.06
	RP-gap	DP (%p)	12.89	8.36	13.42	−0.60	31.51
		ACCT (%p)	11.60	8.95	11.10	−2.57	31.45
		FIN_ACCT (%p)	11.10	8.73	9.54	−1.74	31.80
		CARD (%p)	12.89	8.13	12.66	−1.75	31.97
		SAVING (%p)	13.12	7.24	12.99	−5.16	33.39
		BORROW (%p)	9.71	7.16	8.80	−4.88	34.62
Control variablest		log(GDP)	9.08	1.30	9.11	6.08	11.44
		GROWTH (%)	3.64	2.40	3.51	−6.30	10.24
		log(POP)	2.78	1.48	2.72	−0.34	7.23
		GE (%)	60.44	24.62	58.90	9.85	99.75
		ROL (%)	58.08	26.11	56.50	11.30	99.75

presents descriptive statistics or the main variables in our sample. The financial inclusion index (FII) has a mean of 0.595. In contrast, our key explanatory variable, the abnormal financial inclusion index (AbFII), has a mean of zero by construction, as it is derived as the residual from a regression of FII on GDP per capita. Gender inequality (G-gap) is the percentage point difference between males and females for each financial inclusion indicator in a given country-year. Similarly, rich-poor inequality (RP-gap) captures the percentage point difference between the top 60 percent and bottom 40 percent of the income distribution within each country-year. The mean values for both G-gap and RP-gap are positive, indicating that, on average, men and higher-income individuals have greater access to financial services than women and lower-income groups. Notably, the RP-gap tends to be more pronounced than the G-gap: while all G-gaps average single-digit percentage differences, five of the six RP-gaps exceed 10 percentage points on average.

Table 4. List of countries sorted by financial inclusion index in 2017.

Panel A: Sort by FII
Country	FII	AbFII	GDP ($)	POP (in mil.)	G-gap in DP (%p)	RP-gap in DP (%p)
Canada	0.979	0.077	45,129	30.37	1.30	−0.02
Norway	0.979	−0.011	75,497	4.30	−1.81	−0.45
Sweden	0.970	0.038	53,792	8.20	−0.66	1.89
Denmark	0.967	0.023	57,610	4.78	0.35	−0.27
Finland	0.966	0.060	46,412	4.60	1.57	0.72
⋮
Cote d’Ivoire	0.253	−0.123	2076	13.73	0.10	0.14
Malawi	0.248	0.115	500	9.52	0.06	0.18
Mali	0.240	0.027	796	9.37	0.15	0.06
Cambodia	0.173	−0.137	1401	10.82	0.01	0.08
Madagascar	0.121	−0.014	503	14.61	0.03	0.06
Panel B: Sort by AbFII
Country	AbFII	FII	GDP ($)	POP (in mil.)	G-gap in DP (%p)	RP-gap in DP (%p)
Mongolia	0.372	0.848	3708	2.16	−4.61	7.62
Iran, Islamic Rep.	0.316	0.867	5759	60.44	10.71	3.41
Kenya	0.294	0.634	1676	28.99	9.06	21.05
India	0.241	0.607	1958	954.55	12.30	14.84
Uganda	0.207	0.414	766	20.75	14.58	18.34
⋮
Azerbaijan	−0.226	0.268	4147	7.50	3.41	15.56
El Salvador	−0.227	0.260	3986	4.59	11.76	15.05
Argentina	−0.262	0.448	14,613	32.66	−4.77	14.89
Mexico	−0.311	0.324	9434	89.66	7.69	18.84
Panama	−0.317	0.399	15,186	2.92	10.25	22.20

provides additional context by displaying financial inclusion metrics and other characteristics for selected countries. Panel A lists the five countries with the highest and lowest FII values in 2017. As expected, FII is highly correlated with GDP per capita, which raises concerns about its suitability as an independent proxy for financial inclusion. Moreover, both high-FII and low-FII countries tend to show relatively small levels of gender and rich-poor inequality in financial inclusion. Panel B ranks countries by their AbFII values in the same year. Interestingly, the set of countries identified in Panel B does not overlap with those in Panel A, underscoring that AbFII captures a distinct dimension of financial inclusion—namely, financial inclusion performance beyond what would be predicted by economic development. For instance, countries such as Mongolia and Iran appear at the top of the AbFII ranking, reflecting notably strong financial inclusion outcomes relative to their GDP per capita. Countries with high AbFII may have achieved such outcomes through deliberate policy interventions or institutional innovations to promote financial inclusion, even without high economic development. Table 4 thus serves as a starting point for identifying such countries.

Table 5. Correlation matrix.

	AbFII	FII	G-Gap in DP	RP-Gap in DP	G-Gap in ACCT	RP-Gap in ACCT	log(GDP)
FIIt	0.488 *	-
G-gap in DPt+3	−0.191 *	−0.427 *	-
RP-gap in DPt+3	−0.320 *	−0.625 *	0.376 *	-
G-gap in ACCTt+3	−0.274 *	−0.485 *	0.914 *	0.396 *	-
RP-gap in ACCTt+3	−0.432 *	−0.728 *	0.357 *	0.921 *	0.452 *	-
log(GDP)t	0.002	0.866 *	−0.385 *	−0.530 *	−0.403 *	−0.578 *	-
log(POP)t	0.040	−0.115*	0.086	0.007	0.072	0.040	−0.156 *

Correlation coefficients significant at the 5% level are highlighted with a * sign.

presents Pearson correlation coefficients among selected variables. As expected, AbFII is positively and significantly correlated with FII. In contrast, the correlation between AbFII and the log of GDP per capita is virtually zero (0.002), confirming that AbFII is orthogonal to GDP and thus free from the influence of a country’s economic development level. FII and AbFII exhibit negative correlations with G-gap and RP-gap in financial inclusion metrics. Notably, the correlation between FII and the inequality measures is stronger than that of AbFII. This is primarily because FII is highly correlated with GDP per capita (correlation coefficient = 0.866), and GDP itself is strongly negatively correlated with both G-gap and RP-gap. These results reinforce the validity of using AbFII to isolate the net effect of financial inclusion on inequality, distinct from broader economic factors such as national income levels.

6.2. Univariate Analysis

This section examines whether financial inclusion inequality differs systematically according to a country’s income level or overall inequality. We also explore whether AbFII is associated with improved financial inclusion inequality in a univariate setting. In

Table 6. Mean difference of financial inclusion gaps by dichotomous income and inequality groups.

Panel A: Gender inequality (G-gap) in financial inclusion metrics
	High GDP (n = 100)	Low GDP (n = 100)	t-stat	Good GII (n = 100)	Poor GII (n = 100)	t-stat
G-gap in DP (%p)	3.35	7.93	(4.84)	2.73	8.56	(6.41)
G-gap in ACCT (%p)	2.78	7.64	(5.17)	2.20	8.22	(6.64)
G-gap in FIN_ACCT (%p)	2.79	7.39	(4.98)	2.18	7.99	(6.53)
G-gap in CARD (%p)	3.54	7.79	(4.66)	2.76	8.57	(6.69)
G-gap in SAVING (%p)	3.87	4.25	(0.63)	3.43	4.69	(2.13)
G-gap in BORROW (%p)	5.32	3.09	(−3.15)	4.47	3.93	(−0.75)
Panel B: Rich-poor inequality (RP-gap) in financial inclusion metrics
	High GDP (n = 100)	Low GDP (n = 100)	t-stat	Good Gini (n = 77)	Poor Gini (n = 80)	t-stat
RP-gap in DP (%p)	8.75	17.02	(8.02)	8.05	17.00	(7.46)
RP-gap in ACCT (%p)	6.80	16.40	(8.99)	6.55	15.39	(6.80)
RP-gap in FIN_ACCT (%p)	6.82	15.38	(7.94)	6.38	15.00	(6.78)
RP-gap in CARD (%p)	10.03	15.75	(5.30)	8.57	17.39	(7.85)
RP-gap in SAVING (%p)	15.30	10.94	(−4.46)	14.31	13.54	(−0.67)
RP-gap in BORROW (%p)	12.67	6.75	(−6.40)	10.05	10.56	(0.44)

, we divide the sample into two groups based on the level of GDP per capita and overall inequality levels for each year and compare the mean values of financial inclusion inequality across the groups. Specifically, Panel A focuses on gender inequality (G-gap) in financial inclusion, while Panel B examines income-group inequality (RP-gap). In both panels, countries are further classified by their overall inequality indicators: Panel A uses the UNDP’s Gender Inequality Index (GII), and Panel B uses the Gini coefficient as the income inequality measure. It is worth noting that the sample size in Panel B is somewhat reduced due to missing Gini data for some country-year observations.

Results in Panel A show that gender inequality in financial inclusion is generally more pronounced in low-income countries and countries with higher overall gender inequality. The mean differences between the two groups are statistically significant for DP, a proxy for technology, and ACCT, FIN_ACCT, and CARD, proxies for access. These results suggest that women are disadvantaged in accessing financial services and technology-based financial tools such as digital payments. In contrast, no significant differences are found for savings (SAVING) and borrowing (BORROW), which are categorized as usage dimensions. This suggests that inequalities related to the usage of financial services may behave differently from those related to access and technology, a pattern we further investigate in the multivariate analysis.

Panel B displays a similar pattern for income-group inequality. The RP-gap is consistently larger in low-income countries and countries with high income inequality (high Gini coefficient). The results from both panels imply that countries with lower GII or higher Gini tend to exhibit larger financial inclusion gaps. This motivates our later analysis, where we test whether the effects of AbFII differ depending on a country’s existing level of gender or income inequality.

To what extent can a country’s abnormal financial inclusion alleviate inequality in financial access? To visually address this question, Figure 4 plots changes in gender inequality in digital payments (DP) between 2017 and 2021 for a set of countries. The X-axis and Y-axis represent the G-gap in DP in 2017 and 2021, respectively, with both axes reversed so that movement to the right or upward reflects an improvement in gender equality. Countries located above the diagonal line experienced an improvement in gender inequality in digital payments between 2017 and 2021, while those below the line saw a deterioration. The size of each bubble reflects the country’s AbFII level in 2017.

Consistent with our expectation, countries with larger bubbles (higher AbFII) tend to exhibit greater improvements in gender equality in digital payments. For example, Thailand, which had a high AbFII in 2017, saw its G-gap in DP drop substantially from 12.4% in 2017 to 0.3% in 2021. This pattern mirrors Thailand’s policy in reality. Thailand launched the PromptPay system in 2016, followed by an expansion under the National e-Payment Master Plan by the Bank of Thailand. PromptPay allowed individuals to make real-time transfers using mobile phone numbers or national ID numbers as proxies for bank accounts, significantly reducing transaction costs and improving accessibility. As Yakean (2020) documents, PromptPay became a cornerstone of Thailand’s push towards a cashless society, particularly benefiting low-income and underserved populations. This real-world development is reflected in our finding that Thailand’s abnormal financial inclusion coincided with a dramatic reduction in the gender gap in digital payments.

In contrast, countries with smaller bubbles, indicating lower levels of AbFII, generally experienced less improvement or even deterioration. This pattern provides intuitive visual evidence that AbFII is positively associated with improved financial inclusion equality, particularly in digital payments. However, as these are univariate results, the following section presents multivariate analyses controlling for other country characteristics to validate the relationship further.

6.3. Baseline Multivariate Analysis

Table 7. Abnormal financial inclusion index (AbFII) and inequality in financial inclusion.

Panel A: Gender inequality (G-gap) in financial inclusion metrics
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	G-gap in DP	G-gap in DP	G-gap in ACCT	G-gap in ACCT	G-gap in FIN_ACCT	G-gap in FIN_ACCT	G-gap in CARD	G-gap in CARD	G-gap in SAVING	G-gap in SAVING	G-gap in BORROW	G-gap in BORROW
AbFII	−0.086 **	−0.065 *	−0.124 ***	−0.088 ***	−0.117 ***	−0.073 **	−0.061 *	−0.019	−0.028	−0.013	−0.011	0.016
	(−2.28)	(−1.69)	(−3.80)	(−2.65)	(−3.21)	(−1.98)	(−1.66)	(−0.52)	(−1.10)	(−0.47)	(−0.37)	(0.48)
log(GDP)	−0.009	−0.009	−0.010	−0.016	−0.009	−0.015	−0.006	0.003	0.003	0.000	0.010	0.006
	(−0.95)	(−0.74)	(−1.20)	(−1.60)	(−0.99)	(−1.34)	(−0.61)	(0.25)	(0.48)	(0.01)	(1.27)	(0.63)
GROWTH	0.002	0.005 **	0.000	0.004	0.000	0.004 *	−0.001	0.001	0.000	0.002	0.000	0.002
	(0.99)	(1.98)	(0.03)	(1.64)	(−0.03)	(1.67)	(−0.23)	(0.53)	(−0.14)	(1.00)	(0.14)	(1.13)
log(POP)	0.002	0.002	0.001	0.003	0.002	0.005	0.006 *	0.005 *	0.004 *	0.005 **	−0.002	0.000
	(0.47)	(0.44)	(0.47)	(1.04)	(0.58)	(1.45)	(1.84)	(1.65)	(1.79)	(2.11)	(−0.85)	(0.00)
GE	−0.001	0.000	−0.001 *	0.000	−0.001	0.000	0.000	0.000	0.000	0.000	0.000	0.001
	(−0.96)	(0.49)	(−1.91)	(−0.12)	(−1.61)	(0.09)	(−0.67)	(0.16)	(−0.83)	(0.92)	(−0.12)	(0.85)
ROL	0.000	−0.001	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
	(0.12)	(−0.95)	(1.16)	(0.06)	(0.91)	(0.05)	(−0.47)	(−0.22)	(0.43)	(−0.88)	(−0.22)	(−0.27)
Year FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Region FE		Y		Y		Y		Y		Y		Y
Adj. R2	0.169	0.271	0.256	0.351	0.198	0.305	0.141	0.285	0.000	0.067	0.006	0.026
N	200	200	200	200	200	200	200	200	200	200	200	200
Panel B: Rich-poor inequality (RP-gap) in financial inclusion metrics
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
	RP-gap in DP	RP-gap in DP	RP-gap in ACCT	RP-gap in ACCT	RP-gap in FIN_ACCT	RP-gap in FIN_ACCT	RP-gap in CARD	RP-gap in CARD	RP-gap in SAVING	RP-gap in SAVING	RP-gap in BORROW	RP-gap in BORROW
AbFII	−0.194 ***	−0.165 ***	−0.276 ***	−0.236 ***	−0.260 ***	−0.206 ***	−0.137 ***	−0.085 **	0.136 ***	0.140 ***	0.016	0.064 *
	(−5.00)	(−4.16)	(−7.98)	(−6.63)	(−7.19)	(−5.52)	(−3.16)	(−1.96)	(4.04)	(3.85)	(0.44)	(1.65)
log(GDP)	−0.037 ***	−0.020 *	−0.041 ***	−0.033 ***	−0.034 ***	−0.028 ***	−0.017	−0.012	0.016 *	0.026 **	0.029 ***	0.024 **
	(−3.94)	(−1.84)	(−4.93)	(−3.37)	(−3.87)	(−2.69)	(−1.60)	(−0.96)	(1.87)	(2.37)	(3.17)	(2.19)
GROWTH	0.000	0.000	−0.001	0.000	0.000	0.001	0.000	−0.002	0.004	0.002	0.001	0.002
	(0.03)	(−0.16)	(−0.34)	(0.21)	(−0.03)	(0.50)	(−0.06)	(−0.59)	(1.61)	(0.80)	(0.45)	(0.75)
log(POP)	−0.005	−0.004	−0.003	0.000	−0.003	0.000	0.002	0.002	0.003	0.000	0.000	0.002
	(−1.49)	(−1.16)	(−0.99)	(−0.04)	(−1.05)	(−0.02)	(0.44)	(0.49)	(0.80)	(−0.03)	(0.01)	(0.55)
GE	0.001	0.001	0.001	0.001	0.001 *	0.001 *	0.001	0.001	0.001	0.001	0.000	0.000
	(1.28)	(1.16)	(1.06)	(1.34)	(1.91)	(1.87)	(1.61)	(0.97)	(1.52)	(0.81)	(−0.28)	(0.19)
ROL	−0.001	0.000	−0.001	0.000	−0.001 ***	0.000	−0.001 ***	0.000	0.000	0.000	0.000	0.001
	(−1.63)	(−0.31)	(−1.48)	(−0.47)	(−2.79)	(−0.90)	(−2.63)	(0.02)	(−1.04)	(−0.44)	(0.46)	(1.20)
Year FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Region FE		Y		Y		Y		Y		Y		Y
Adj. R2	0.387	0.450	0.550	0.591	0.476	0.526	0.188	0.291	0.271	0.275	0.251	0.300
N	200	200	200	200	200	200	200	200	200	200	200	200

t-values are shown in parentheses. ***, **, and * indicates significance at 1%, 5%, and 10% levels, respectively.

reports the estimation results of Equation (2), where we examine the relationship between AbFII and financial inclusion inequality across multiple dimensions. The dependent variables are the G-gap and RP-gap for all six financial inclusion metrics. Panel A presents the results estimated with G-gaps as dependent variables, and Panel B shows those estimated with RP-gaps as dependent variables. In each panel, odd-numbered columns include only year-fixed effects, and even-numbered columns additionally control for region-fixed effects.

Columns (1) and (2) of Panel A show that AbFII has a negative and statistically significant effect on the G-gap in digital payments (DP). In other words, higher abnormal financial inclusion is associated with improved gender equality in digital payments after controlling for the country’s economic strength. Given that the standard deviation of AbFII is 0.131 (see Table 2), the estimated coefficient implies that a standard deviation increase in AbFII reduces the G-gap in DP by about 1.1 percentage points (0.131 × 0.086 = 0.011 or 1.1%p), which is economically meaningful relative to the sample mean G-gap of 5.64 percentage points (See Table 3).

In columns (3) through (8), we observe similar patterns: AbFII is negatively and significantly associated with gender inequality in account ownership (ACCT), financial institution account ownership (FIN_ACCT), and debit or credit card ownership (CARD). These results suggest that AbFII has a positive effect on improving gender equality in various financial inclusion metrics. The estimated effects are again sizable, with a one standard deviation increase in AbFII reducing the G-gap by about 1.3 percentage points for ACCT, 1.2 percentage points for FIN_ACCT, and 0.5 percentage points for CARD. By contrast, the results in columns (9) through (12) show that the effect of AbFII on the G-gap in saving at financial institutions (SAVING) and borrowing from financial institutions (BORROW) is insignificant. A plausible explanation is that, although saving and borrowing are essential aspects of financial inclusion, such decisions are more directly influenced by individuals’ financial capacity rather than mere access to financial services.

We find similar patterns when examining the effect of AbFII on the RP-gap in financial inclusion, as reported in Panel B. In columns (1) through (8), the estimated coefficients of AbFII are consistently negative and statistically significant, indicating that higher abnormal financial inclusion is associated with a smaller RP-gap across four key financial inclusion metrics: DP, ACCT, FIN_ACCT, and CARD. Notably, the magnitude of the coefficients is larger in Panel B than in Panel A, suggesting that AbFII may have a stronger impact on reducing rich-poor inequality than gender inequality. This finding aligns with the observation that RP-gaps are substantially larger than G-gaps in our sample, implying that there is more room for improvement in rich-poor disparities.

These results suggest that a country’s disproportionate investment in financial inclusion beyond what its economic development alone would predict contributes meaningfully to reducing gender and income-group inequalities in financial access. This provides novel evidence that addressing financial inclusion disparities requires more than just standard investment levels; additional and possibly targeted country-level efforts are needed. Importantly, these findings remain robust when controlling for a comprehensive set of country-level characteristics that could otherwise confound the relationship between AbFII and financial inclusion inequality.

6.4. Heterogeneity in the Impact of Abnormal FII on Inequality in Financial Inclusion

In this section, we investigate whether the impact of AbFII on improving financial inclusion inequality varies depending on a country’s overall level of gender or income inequality. Specifically, we examine whether the positive effect of AbFII on reducing the G-gap is more pronounced in countries with higher levels of overall gender inequality. Similarly, we test whether the effect of AbFII on reducing the RP-gap is stronger in countries with higher income inequality. We measure overall gender inequality using the Gender Inequality Index (GII) published by the United Nations Development Program (UNDP), which captures three key dimensions: reproductive health, empowerment, and labor market participation. We use country-year data on the Gini coefficient obtained from the World Bank for income inequality. Intuitively, one may expect that financial inclusion could have a larger effect in reducing inequality in financial access in countries with more severe overall inequality, as these countries have greater room for improvement. On the other hand, it is also plausible that a highly unequal environment could hinder the equalizing effect of financial inclusion. Therefore, the extent to which the AbFII effect varies across countries is an empirical question. If AbFII proves to have a stronger effect in countries with higher inequality, this would provide valuable insights for policymakers.

To empirically test this, we divide countries into two groups based on their yearly GII and Gini index levels. Specifically, we define Good GII as an indicator variable equal to one if a country’s GII is below the annual median (i.e., better gender equality) and Poor GII as one above the median. Likewise, Good Gini equals one for countries below the median Gini coefficient (lower income inequality), and Poor Gini equals one for countries above the median.

Table 8. Heterogeneous effects of abnormal FII by countries’ overall gender and rich-poor inequality.

Panel A: Gender inequality (G-gap) in financial inclusion metrics
	(1)	(2)	(3)	(4)	(5)	(6)
	G-gap in DP	G-gap in ACCT	G-gap in FIN_ACCT	G-gap in CARD	G-gap in SAVING	G-gap in BORROW
AbFII * Poor GII	−0.080 *	−0.107 ***	−0.085 *	−0.027	0.004	0.039
	(−1.74)	(−2.71)	(−1.94)	(−0.63)	(0.10)	(1.01)
AbFII * Good GII	−0.003	−0.016	−0.002	0.048	−0.046	−0.012
	(−0.04)	(−0.26)	(−0.03)	(0.69)	(−1.00)	(−0.20)
log(GDP)	−0.007	−0.014	−0.013	0.005	0.000	0.007
	(−0.61)	(−1.43)	(−1.16)	(0.43)	(−0.05)	(0.68)
GROWTH	0.005 **	0.004 *	0.004 *	0.001	0.001	0.002
	(2.00)	(1.68)	(1.68)	(0.53)	(0.96)	(1.02)
log(POP)	0.002	0.003	0.005	0.005 *	0.005 **	0.000
	(0.43)	(1.03)	(1.48)	(1.69)	(2.09)	(−0.01)
GE	0.000	0.000	0.000	0.000	0.000	0.001
	(0.43)	(−0.20)	(0.03)	(0.11)	(0.96)	(0.91)
ROL	0.000	0.000	0.000	0.000	0.000	0.000
	(−0.91)	(0.10)	(0.11)	(−0.16)	(−0.92)	(−0.27)
Good GII	−0.010	−0.009	−0.018	−0.022	0.001	−0.018
	(−0.59)	(−0.60)	(−1.14)	(−1.38)	(0.08)	(−1.27)
Year FE	Y	Y	Y	Y	Y	Y
Region FE	Y	Y	Y	Y	Y	Y
Adj. R2	0.268	0.351	0.307	0.288	0.061	0.026
N	200	200	200	200	200	200
Panel B: Rich-poor inequality (RP-gap) in financial inclusion metrics
	(1)	(2)	(3)	(4)	(5)	(6)
	RP-gap in DP	RP-gap in ACCT	RP-gap in FIN_ACCT	RP-gap in CARD	RP-gap in SAVING	RP-gap in BORROW
AbFII * Poor Gini	−0.178 ***	−0.267 ***	−0.252 ***	−0.138 **	0.070	0.030
	(−3.21)	(−5.41)	(−4.90)	(−2.27)	(1.22)	(0.50)
AbFII * Good Gini	−0.200 ***	−0.233 ***	−0.209 ***	−0.072	0.157 **	0.140 **
	(−3.12)	(−4.02)	(−3.48)	(−1.02)	(2.34)	(1.98)
log(GDP)	−0.042 ***	−0.052 ***	−0.044 ***	−0.035 **	0.021	0.018
	(−3.33)	(−4.66)	(−3.81)	(−2.57)	(1.53)	(1.31)
GROWTH	0.002	0.002	0.003	0.000	0.002	0.003
	(0.75)	(0.96)	(1.28)	(0.03)	(0.76)	(0.88)
log(POP)	−0.005	−0.001	−0.001	0.000	−0.001	0.004
	(−1.46)	(−0.31)	(−0.43)	(0.02)	(−0.17)	(0.91)
GE	0.001	0.001	0.001	0.001	0.000	0.000
	(0.68)	(1.08)	(1.27)	(0.74)	(0.32)	(−0.37)
ROL	0.000	0.000	0.000	0.000	0.000	0.001 *
	(0.53)	(0.52)	(0.21)	(0.50)	(−0.20)	(1.75)
Good Gini	−0.051 ***	−0.038 ***	−0.039 ***	−0.066 ***	−0.014	−0.034 **
	(−4.19)	(−3.53)	(−3.46)	(−4.94)	(−1.13)	(−2.56)
Year FE	Y	Y	Y	Y	Y	Y
Region FE	Y	Y	Y	Y	Y	Y
Adj. R2	0.574	0.678	0.633	0.433	0.189	0.246
N	157	157	157	157	157	157

t-values are shown in parentheses. ***, **, and * indicates significance at 1%, 5%, and 10% levels, respectively.

reports the results from the baseline regressions augmented with interaction terms between AbFII and the inequality indicators (Poor GII and Poor Gini). In this specification, the interaction terms directly capture whether the effect of AbFII differs across low-inequality and high-inequality countries. As in previous tables, Panel A focuses on gender inequality (G-gap) and Panel B on income inequality (RP-gap).

Panel A shows that the interaction terms between AbFII and Poor GII are consistently negative and statistically significant, whereas those with Good GII are insignificant. This result indicates that AbFII contributes more effectively to reducing gender inequality in financial inclusion in countries with higher overall gender inequality. In contrast, for countries with better overall gender equality (Good GII), AbFII does not significantly affect the G-gap. However, Panel B shows a different pattern regarding income inequality. The effects of AbFII on RP-gap reduction are similar between Poor Gini and Good Gini countries across most financial inclusion metrics (DP, ACCT, and FIN_ACCT). Only in the case of CARD does AbFII show a significantly stronger effect in countries with higher income inequality.

Overall, our findings suggest that AbFII has a stronger impact on the G-gap in financial inclusion in countries with greater baseline gender inequality. However, the effect on rich-poor inequality appears less sensitive to a country’s level of overall income inequality. This pattern may arise because, in countries with more severe gender inequality, improving access through financial inclusion often directly addresses structural or institutional barriers women face, amplifying the equalizing effect of AbFII. In contrast, income inequality may not exhibit a similarly clear pattern, as a country’s overall income distribution is often deeply rooted in its economic structure, labor market dynamics, and demographic composition (Brei et al., 2023; Kling et al., 2022). These structural factors may limit how improved access to financial services can immediately narrow rich-poor disparities in financial inclusion.7

6.5. Does Abnormal FII Improve Overall Gender Inequality?

The primary objective of this study is to examine the relationship between a country’s abnormal financial inclusion and inequality in financial inclusion. In this section, however, we extend our analysis to explore whether abnormal financial inclusion reduces broader societal inequality, particularly gender inequality. Since improving financial inclusion for disadvantaged groups can enhance their participation in various aspects of society, it is plausible to expect that reducing inequality in financial inclusion may, in turn, improve overall gender inequality.

We use the Gender Inequality Index (GII) published by the United Nations Development Programme (UNDP) to measure overall gender inequality. The GII reflects three key dimensions: reproductive health, empowerment, and labor market participation. It comprises several specific subcomponents, including the maternal mortality ratio, adolescent birth rate, the proportion of the population with higher education, the share of parliamentary seats held by women, and the female labor force participation rate.

Table 9. Abnormal financial inclusion index (AbFII) and overall gender inequality.

	(1)	(2)	(3)	(4)	(5)	(6)
	Gender Inequality Index (t + 1)	Maternal Mortality %	Adolescent Birth %	Higher Edu. Inequality	% Seats in Parliament	Labor Participation Inequality
AbFII	−0.140 ***	−77.769 *	−21.079 ***	−1.950	−3.450	−7.306
	(−3.68)	(−1.92)	(−2.75)	(−0.65)	(−0.55)	(−1.57)
log(GDP)	−0.030 ***	−0.200	−6.709 ***	−0.514	−3.764 **	−1.152
	(−3.28)	(−0.02)	(−3.13)	(−0.57)	(−2.14)	(−0.83)
GROWTH	0.004 **	6.686 **	0.340	0.190	−0.487	0.465 *
	(2.42)	(2.52)	(0.80)	(1.14)	(−1.19)	(1.81)
log(POP)	0.008 **	13.446 ***	1.077	0.650 **	0.244	0.199
	(2.47)	(3.77)	(1.62)	(2.46)	(0.45)	(0.49)
GE	−0.002 ***	−2.469 ***	−0.651 ***	0.025	0.267 **	0.107
	(−2.58)	(−3.11)	(−4.52)	(0.44)	(2.28)	(1.23)
ROL	0.000	−0.521	0.160	−0.057	0.041	−0.123 *
	(−0.93)	(−0.90)	(1.45)	(−1.31)	(0.46)	(−1.84)
Year FE	Y	Y	Y	Y	Y	Y
Region FE	Y	Y	Y	Y	Y	Y
Adj. R2	0.917	0.807	0.877	0.227	0.248	0.560
N	200	200	200	200	200	200

t-values are shown in parentheses. ***, **, and * indicates significance at 1%, 5%, and 10% levels, respectively.

presents regression results where the dependent variable is the GII in column (1) and its five subcomponents in columns (2) through (6). Consistent with our primary analysis, we use the year t + 3 value of the dependent variables and the year t value of the explanatory variables. All regressions control for the same set of country-level characteristics and year and region-fixed effects.

Column (1) shows the result consistent with our expectation. The coefficient of AbFII is negative and statistically significant, suggesting that higher abnormal financial inclusion is associated with lower gender inequality in subsequent years. Given that a lower GII indicates greater gender equality, the coefficient of −0.140 suggests that a one standard deviation increase in AbFII reduces the GII by approximately 0.02, which translates into an improvement of about seven positions in the country’s gender inequality ranking, which is a meaningful improvement in economic terms.

Next, Columns (2) to (6) further explore which aspects of gender inequality are most affected by AbFII. The results indicate that improvements in reproductive health primarily drive the reduction in gender inequality, as the coefficients for maternal mortality (column 2) and adolescent birth rate (column 3) are both negative and statistically significant. However, the effects of AbFII on the other dimensions of gender inequality—education, political empowerment, and labor force participation—are not statistically significant. This does not necessarily imply that AbFII does not affect these components. Instead, these results may be due to measurement limitations. For example, the UNDP defines higher education attainment based on individuals aged 25 or older, which may be less sensitive to changes within a three-year horizon. Similarly, electoral cycles and institutional factors often influence female representation in parliament, which may limit the short-term responsiveness to financial inclusion improvements.

Overall, our findings indicate that abnormal financial inclusion is associated with improvements in broader gender inequality, especially in areas where financial access may immediately influence outcomes, such as reproductive health. These results are consistent with the idea that improving access to financial services for disadvantaged groups can help narrow gender gaps across various socio-economic outcomes. Nevertheless, given many other determinants of overall gender inequality beyond financial inclusion, we remain cautious in making strong causal claims based solely on these results.

6.6. Robustness Checks

In this section, we conduct a robustness check to reinforce the credibility of our findings. A key identification challenge in our analysis is establishing that AbFII causally reduces gender and rich-poor inequalities in financial inclusion. Although we believe that endogeneity concerns such as reverse causality or simultaneity are unlikely given the timing of our dependent variable (the G-gap or RP-gap in year t + 3) and the primary explanatory variable (AbFII in year t), we cannot fully rule out the possibility that omitted variables may bias our estimates.8

We employ a two-stage least squares (2SLS) estimation strategy to address this concern. Specifically, we use the number of fixed-broadband internet subscribers relative to the country’s population as an instrumental variable (IV) to capture the exogenous variation in AbFII. Data for the instrument is obtained from the International Telecommunication Union (ITU), which annually reports country-level data on the number of subscribers with access to high-speed public internet (TCP/IP connections) at downstream speeds of 256 kbit/s or higher. The number of subscribers varies significantly across countries, ranging from 46.3% of the population in Switzerland to only 0.1% in Nigeria as of 2017.

We use broadband penetration as an instrument for AbFII because it is strongly associated with financial infrastructure yet plausibly unrelated to group-specific financial access disparities. We argue that broadband penetration satisfies a valid instrument’s relevance and exclusion conditions. First, countries with better network infrastructure are more likely to facilitate additional investments in fintech, positively influencing AbFII. Second, network infrastructure alone is unlikely to affect financial inclusion inequality directly; instead, its effect on inequality would operate solely through its influence on financial inclusion, not directly on the G-gap or RP-gap. Therefore, broadband penetration meets the exclusion restriction, as it indirectly influences group-level disparities in financial inclusion through its effect on national inclusion efforts rather than through any direct channel.

Table 10. Two-stage least squares (2SLS) analysis.

Panel A: Gender inequality (G-gap) in financial inclusion metrics
	First Stage	Second Stage
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	AbFII	G-gap inDP	G-gap in ACCT	G-gap in FIN_ACCT	G-gap in CARD	G-gap in SAVING	G-gap in BORROW
AbFII		−0.632 **	−0.428 **	−0.507 **	−0.558 **	−0.194	−0.071
		(−2.35)	(−2.12)	(−2.14)	(−2.13)	(−1.43)	(−0.44)
log(GDP)	−0.063 ***	−0.030	−0.027 *	−0.029 *	−0.016	−0.006	0.003
	(−3.12)	(−1.55)	(−1.89)	(−1.75)	(−0.85)	(−0.62)	(0.30)
GROWTH	−0.006	0.001	0.001	0.001	−0.002	0.000	0.002
	(−1.37)	(0.26)	(0.46)	(0.36)	(−0.59)	(0.12)	(0.78)
log(POP)	−0.001	0.001	0.002	0.003	0.004	0.004 *	0.000
	(−0.21)	(0.17)	(0.55)	(0.81)	(0.81)	(1.77)	(−0.11)
GE	0.002 *	0.001	0.000	0.001	0.001	0.001	0.001
	(1.77)	(1.26)	(0.59)	(0.81)	(0.95)	(1.32)	(0.99)
ROL	−0.001	−0.001	0.000	0.000	0.000	0.000	0.000
	(−0.61)	(−0.83)	(−0.21)	(−0.24)	(−0.45)	(−0.94)	(−0.36)
INTERNET	0.004 ***
	(3.00)
Year FE	Y	Y	Y	Y	Y	Y	Y
Region FE	Y	Y	Y	Y	Y	Y	Y
Adj. R2	0.350
F statistics	8.97
F statistics p-value	0.00
N	199	199	199	199	199	199	199
Panel B: Rich-poor inequality (RP-gap) in financial inclusion metrics
	First Stage	Second Stage
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	AbFII	RP-gap in DP	RP-gap in ACCT	RP-gap in FIN_ACCT	RP-gap in CARD	RP-gap in SAVING	RP-gap in BORROW
AbFII		−0.407 *	−0.447 **	−0.498 **	−0.483 *	0.161	−0.197
		(−1.94)	(−2.41)	(−2.46)	(−1.91)	(0.88)	(−0.94)
log(GDP)	−0.063 ***	−0.027 *	−0.042 ***	−0.040 ***	−0.027	0.029 **	0.014
	(−3.12)	(−1.91)	(−3.23)	(−2.79)	(−1.56)	(2.33)	(0.95)
GROWTH	−0.006	−0.002	−0.001	−0.001	−0.004	0.002	0.000
	(−1.37)	(−0.73)	(−0.30)	(−0.19)	(−1.14)	(0.69)	(0.10)
log(POP)	−0.001	−0.005	−0.001	−0.001	0.001	−0.001	0.001
	(−0.21)	(−1.39)	(−0.18)	(−0.17)	(0.19)	(−0.20)	(0.34)
GE	0.002 *	0.001	0.001	0.002 **	0.002	0.000	0.001
	(1.77)	(1.38)	(1.63)	(2.16)	(1.41)	(0.66)	(0.67)
ROL	−0.001	0.000	0.000	−0.001	0.000	0.000	0.001
	(−0.61)	(−0.40)	(−0.54)	(−0.99)	(−0.09)	(−0.56)	(1.03)
INTERNET	0.004 ***
	(3.00)
Year FE	Y	Y	Y	Y	Y	Y	Y
Region FE	Y	Y	Y	Y	Y	Y	Y
R2	0.350
F statistics	8.97
F statistics p-value	0.00
N	199	199	199	199	199	199	199

t-values are shown in parentheses. ***, **, and * indicates significance at 1%, 5%, and 10% levels, respectively.

shows the 2SLS results. We again replicate the baseline model from Table 6 to examine the effect of AbFII on the G-gap and RP-gap in various financial inclusion measures, using the instrumented ABFII in the second-stage regressions. Panel A reports results for gender inequality, while Panel B focuses on rich-poor inequality. Column (1) presents the first-stage regression in both panels, where ABFII is the dependent variable. Columns (2) through (7) report the second-stage results, where the dependent variables are the same inequality measures used in Table 7.

Consistent with the relevance condition, the first-stage results in column (1) show that broadband penetration is positively and significantly related to AbFII. The F-statistic is 8.97, slightly below the conventional threshold of 10 but still statistically significant, suggesting the instrument is informative. Overall, the first-stage results confirm the relevance of the instrumental variable. Next, in both panels, the second-stage estimates in columns (2) through (7) are qualitatively similar to our baseline OLS results and, in many cases, show larger coefficient magnitudes. This pattern suggests that any potential endogeneity may have led to attenuation bias in our baseline estimates, further reinforcing the robustness of our main findings. However, we also note that the instrument used in the 2SLS regressions may not be sufficiently strong, and thus, the results should be interpreted with appropriate caution regarding causal inference.

7. Conclusions

The United Nations Development Program (UNDP) adopted 17 Sustainable Development Goals, including equality in financial inclusion across genders (Goal 5) and different income groups within and among countries (Goal 10). Goal 5 is to achieve gender equality and empower all women by undertaking reforms to give them equal rights to economic resources and access to financial services. Goal 10 is to achieve the reduction of inequality within and among countries. Our analysis reveals a promising trend in the overall levels of financial inclusion globally. However, when we examine inequality within financial inclusion, clear challenges remain: gender and rich-poor disparities in financial access have not improved as expected and, in some cases, have even worsened.

This study examines whether a country’s unexpectedly high investment in financial inclusion reduces gender and income-group disparities in financial access. Using the World Bank’s Global Findex data from 100 countries across 2014, 2017, and 2021, we develop a novel abnormal financial inclusion index (AbFII), which captures the component of financial inclusion that exceeds what would be expected given a country’s GDP per capita. Our empirical findings show that countries with higher AbFII experience significant reductions in gender and rich-poor inequalities in financial inclusion, particularly in digital payments, account ownership, and access to financial institutions. The effects are more pronounced for gender inequality in countries with greater baseline gender inequality, suggesting that additional efforts in financial inclusion may be especially effective where structural barriers are more severe. In contrast, the impact on income-group inequality appears less sensitive to the country’s level of overall income inequality.

This paper contributes to the literature by providing the first cross-country evidence that abnormal levels of financial inclusion beyond those predicted by economic development play a crucial role in mitigating inequality in financial access. Our findings also have important policy implications: promoting financial inclusion alone may not be sufficient; disproportionate and targeted efforts may be necessary, especially in countries with severe gender or income disparities.

Nonetheless, our analysis has several limitations. While we interpret AbFII as reflecting a country’s abnormal or disproportionate investment in financial inclusion, it is essential to note that we do not directly observe actual financial inclusion investment activities or expenditures. Instead, AbFII is derived from survey-based indicators in the Global Findex database, which captures financial inclusion outcomes rather than inputs. Furthermore, the data availability limits our analysis to 100 countries and three survey years. Future research could address these limitations by incorporating more direct measures of financial inclusion investment and exploring how such investments translate into broader social and economic outcomes.

Also, while our analysis focuses on national-level outcomes related to gender and income-group disparities, we acknowledge that in a corporate context, an essential consideration for some stakeholders is whether financial inclusion policies are viable and relevant in enhancing firm value. Since companies exist to deliver economically meaningful products and services, provide employment, and generate profits for shareholders, evaluating such policies through their contribution to those objectives could be a meaningful direction for future research.