The Nonlinear Effect of Financial Development on Income Inequality: New Evidence from a Multi-Dimensional Analysis

Abstract

Over the past three decades, rising income inequality has undermined economic performance and posed challenges for policymakers, highlighting the need to identify its underlying drivers to design effective policy responses. Financial development is often considered a potential driver of inequality, yet the theoretical and empirical literature on how financial development affects inequality remains inconclusive. Moreover, prior studies have primarily relied on traditional indicators, which do not comprehensively reflect the multidimensional nature of financial development. To address these gaps, we provide the first study to employ the IMF’s Financial Development Index and all its sub-indices within both fixed-effects and system GMM frameworks to examine whether financial development and its dimensions exhibit a nonlinear relationship with income inequality. Unlike traditional indicators, these indices offer a more comprehensive view of financial development. Using panel data for 130 countries from 1980 to 2019, we find that financial development and its dimensions—access to financial institutions (financial inclusion) and depth of financial institutions—initially reduce inequality but exacerbate it once their respective thresholds are exceeded. These results are not driven by systemic banking crises. Our study contributes by providing a more comprehensive assessment, demonstrating nonlinear effects, identifying thresholds, and offering policy implications for countries at different income levels.

1. Introduction

Over the past three decades, income inequality has risen sharply, posing serious challenges for policymakers. Tackling inequality is essential—not only to promote fairness and social cohesion but also to reduce socio-political instability and preserve economic stability and growth. Prior research shows that inequality can undermine economic performance by fueling instability and discouraging investment (Alesina & Perotti, 1996). These challenges suggest the importance of identifying the underlying drivers of rising inequality to design effective policy responses.

Theoretical literature suggests that financial development is a potential driver of income inequality. One strand proposes that deeper financial systems may mitigate inequality by improving credit allocation (Banerjee & Newman, 1993; Galor & Zeira, 1993), while another argues that the financial sector can exacerbate inequality by facilitating rent extraction (Gennaioli et al., 2012; Thakor, 2012; Bolton et al., 2016). Some studies also indicate a nonlinear relationship (Tan & Law, 2012; Cihak & Sahay, 2020).

These theoretical predictions have motivated empirical research, but results remain mixed. Some studies find that financial development increases income inequality (Jung & Cha, 2021), whereas others report a mitigating effect (Thornton & Di Tommaso, 2020; Manta et al., 2023). Some studies also highlight nonlinear effects: Tan and Law (2012), Cihak and Sahay (2020), and Brei et al. (2023) find a U-shaped relationship between financial development and inequality.

Prior studies have primarily relied on financial depth—particularly market capitalization to GDP or private credit to GDP—to measure financial development. However, these indicators capture only a subset of the multidimensional nature of financial development. For instance, private credit to GDP overlooks the role of non-bank financial institutions, such as mutual funds, pension funds, and insurance companies. It also does not account for access or efficiency dimensions of financial institutions. Similarly, market capitalization to GDP reflects stock market size but fails to capture stock market liquidity, debt market development, or access and efficiency dimensions of financial markets. To address these limitations, the IMF developed the Financial Development Index, which aggregates nine sub-indices covering depth, access, and efficiency of financial institutions and financial markets (Svirydzenka, 2016). Using this index and some of its sub-indices, Cihak and Sahay (2020) and Brei et al. (2023) apply the system GMM model and find a U-shaped relationship between financial development and income inequality. However, their analysis does not utilize all sub-indices and therefore does not cover all key dimensions of financial development, leaving room for further research.

Furthermore, the existing literature indicates the need to distinguish between financial institution-based and market-based financial development when examining its effect on income inequality. Brei et al. (2023) demonstrate that the financial structure—whether bank-based or market-based—changes the relationship between financial development and inequality. Banks and financial markets differ in accessibility, competitive dynamics, and the types of information they rely on. While banks tend to be more accessible and partially depend on private information, financial markets usually have more restricted participation, trade predominantly using public information, and rely on formal contracting. To differentiate between bank-based and market-based financial development, Brei et al. (2023) employ two proxies of financial depth: bank credit to GDP and stock market capitalization to GDP. However, these measures may overlook key aspects of financial depth, suggesting the need for a more comprehensive analysis.

Moreover, the literature suggests that systemic banking crises might affect the relationship between financial development and income inequality. For example, De Haan and Sturm (2017) and Cihak and Sahay (2020) find that systemic banking crises exacerbate inequality. This is primarily because low-income individuals, who have limited means to hedge against shocks and are more vulnerable to job loss during downturns, are disproportionately affected. However, empirical evidence on whether systemic banking crises drive the effect of financial development on income inequality remains scarce, highlighting another avenue for future research.

Building on these theoretical insights, mixed empirical findings, the limitations of conventional measures in capturing the multidimensional nature of financial development, and the underutilization of the Financial Development Index and its sub-indices in prior studies, we are motivated to examine whether financial development and its dimensions—the depth, efficiency, and access to financial institutions and markets—exhibit a nonlinear relationship with income inequality using the IMF’s Financial Development Index and all its sub-indices within a fixed-effects and system GMM framework. In other words, our research question is as follows: does financial development and its dimensions exhibit a nonlinear relationship with income inequality? We also conduct robustness tests to examine whether these nonlinearities are driven by systemic banking crises.

This study makes five main contributions. First, to the best of our knowledge, this is the first study to incorporate the IMF’s Financial Development Index and all eight of its sub-indices into both fixed-effects and system GMM frameworks to examine the nonlinear effects of financial development and its dimensions on income inequality. Unlike prior studies that primarily relied on indicators such as market capitalization to GDP or private credit to GDP, our approach provides a more comprehensive assessment by accounting for the depth, access, and efficiency of financial institutions and markets.

Second, by analyzing each sub-index separately, this study offers a more nuanced understanding of which specific dimensions of financial development reduce or exacerbate inequality. This disaggregated approach enables policymakers to design more targeted and effective policy interventions.

Third, to the best of our knowledge, this is the first study to combine the systemic banking crisis dummy with the IMF’s Financial Development Index and all of its sub-indices to explore the nonlinear relationship between financial development and income inequality. By using the more recent datasets from L. Laeven and Valencia (2020) and Nguyen et al. (2022), our analysis extends through 2019, offering broader temporal coverage than earlier studies such as Cihak and Sahay (2020) and Brei et al. (2023), which relied on M. L. Laeven and Valencia (2018) and were limited to data up to 2017.

Fourth, this study draws on broader and more up-to-date datasets, enabling more robust conclusions and a deeper understanding of economic changes that earlier research may have overlooked. In contrast, earlier studies such as Cihak and Sahay (2020) and Brei et al. (2023) only cover periods ending in 2015 and 2012, respectively.

Lastly, this study identifies U-shaped effects of financial development, financial depth, and financial access on inequality, with thresholds that help policymakers understand when these factors reduce or worsen inequality, highlighting the need for stage-specific and well-targeted financial policies.

2. Literature Review

Theoretical literature suggests several mechanisms through which financial depth—a key dimension of financial development—may affect income inequality. One strand proposes that deeper financial systems may mitigate inequality. Banerjee and Newman (1993) and Galor and Zeira (1993) argue that greater credit availability enables households to better allocate capital over time. This improved allocation facilitates a broader range of decisions that are independent of inherited wealth, ultimately alleviating inequality. Another strand suggests that the financial sector can exacerbate inequality by facilitating rent extraction. Gennaioli et al. (2012), Thakor (2012), and Bolton et al. (2016) argue that inefficient or harmful financial innovations can raise rent extraction and worsen inequality. Korinek and Kreamer (2014) demonstrate that financial deregulation may increase inequality. Synthesizing these perspectives, Cihak and Sahay (2020) propose that these theories may be complementary: the first group is more applicable to developing countries, while the second group better explains patterns in developed countries. Hence, Cihak and Sahay (2020) suggest a U-shaped relationship between financial depth and inequality. Furthermore, Tan and Law (2012) suggest that financial depth reduces inequality at the early stages of financial development but exacerbates it beyond a certain threshold due to increasing inefficiencies.

Based on these mechanisms, our first hypothesis is that financial depth has a U-shaped relationship with income inequality: financial depth initially reduces inequality by improving credit allocation, enabling households to better allocate capital over time; beyond a certain threshold, further increases in financial depth may increase inequality by fostering rent extraction and amplifying inefficiencies.

Theoretical literature provides insights into the mechanisms through which financial access and inclusion—a key dimension of financial development—affect income inequality.1 Banerjee and Newman (1993), Galor and Zeira (1993), and Aghion and Bolton (1997) argue that credit market imperfections prevent the poor from investing in education or entrepreneurship, thereby reinforcing inequality. Enhancing access to financial services helps alleviate these constraints, facilitates greater investment in human and physical capital, and ultimately reduces inequality. However, Inoue (2024) argues that as financial access and inclusion advance, those who have already benefited from financial services are better positioned to derive further advantages due to their superior social and economic standing, while the financially marginalized may not benefit to the same extent. Moreover, Inoue (2024) suggests that once equal access to basic financial services becomes widespread, government intervention to ensure inclusiveness may decline, and financial institutions may shift toward high value-added services for affluent clients, thereby reinforcing inequality by favoring the already well-off.

Based on these mechanisms, our second hypothesis is that there is a U-shaped effect of financial access on inequality: initial improvements in access reduce inequality by alleviating credit market imperfections and enabling greater investment in human and physical capital, but beyond a threshold, further expansion may worsen inequality because those who have already benefited are better positioned to gain additional advantages, while financially marginalized households benefit less, and financial institutions may shift toward high value-added services favoring affluent clients.

In addition, since efficiency is also a dimension of financial development and empirical studies investigating its effect on inequality are scarce, we explore this effect as well. However, given the lack of theoretical foundations, we do not formulate an explicit hypothesis for this dimension.

In our empirical analysis, we remain agnostic about the specific channels through which financial development affects inequality, concentrating instead on evidence supporting the more fundamental proposition of a nonlinear relationship.

A growing body of empirical literature highlights the nonlinear relationship between financial development and income inequality. Tan and Law (2012) find that financial depth initially reduces inequality but increases it once a certain threshold is exceeded. However, their analysis is limited to 35 developing economies between 1980 and 2000 and does not include either developed economies or a broader set of developing countries. Moreover, they measure financial development using private credit to GDP and market capitalization to GDP—both indicators of financial depth—while overlooking the role of non-bank financial institutions and not accounting for access and efficiency dimensions of financial development. In addition, their analysis relies solely on a system GMM framework without exploring robustness to alternative estimation models, such as a fixed-effects model, nor do they examine whether the results are driven by systemic banking crises.

Furthermore, Cihak and Sahay (2020) use two sub-indices of the IMF’s Financial Development Index for 128 countries (1980–2010) within a system GMM framework and find a U-shaped relationship between financial depth and inequality. However, they do not utilize the IMF’s Financial Development Index and its remaining six sub-indices. Consequently, their analysis is limited to financial depth and excludes the access and efficiency dimensions of financial development, as well as the effect of overall financial development on inequality. Tan and Law (2012), they rely on a system GMM approach, without testing whether the results are robust to alternative estimation models or whether they are driven by systemic banking crises.

A recent study by Brei et al. (2023) employs the IMF’s Financial Development Index and three weighted-average sub-indices from 97 countries between 1989 and 2012. They find that financial development initially reduces inequality but worsens it once a certain threshold is exceeded. They also find that neither financial depth nor financial efficiency has a robust effect on inequality. However, their analysis does not investigate whether a specific dimension of financial development affects inequality differently across financial institutions and markets, indicating the need for a more comprehensive approach that incorporates all sub-indices of the IMF’s Financial Development Index. Like Tan and Law (2012) and Cihak and Sahay (2020), their analysis relies on a system GMM framework without testing robustness under alternative estimation models.

3. Materials and Methods

This study uses panel data for 130 countries from 1980 to 2019. Financial development is measured using the Financial Development Index (FD), sourced from the IMF’s Financial Development Index Database. The index evaluates financial development in terms of depth, efficiency, and access for financial institutions and markets. Additionally, the study incorporates eight sub-indices of FD to examine specific dimensions of financial development: the Financial Institutions Index (FI), Financial Institutions Depth Index (FID), Financial Institutions Efficiency Index (FIE), Financial Institutions Access Index (FIA), Financial Markets Index (FM), Financial Markets Depth Index (FMD), Financial Markets Efficiency Index (FME), and Financial Markets Access Index (FMA), which are described in

Table A1. Description of the sub-indices of the Financial Development Index.

FI	Depth, efficiency, and access within financial institutions
FID	Mutual fund assets, pension fund assets, insurance premiums, and private-sector credit (% of GDP)
FIE	Return on assets, overhead costs to total assets, and net interest margin
FIA	Total number of bank branches and ATMs per 100,000 adults
FM	Depth, efficiency, and access within financial markets
FMD	Total debt securities, stock market capitalization, and stocks traded (% of GDP)
FME	Stock market turnover ratio
FMA	Total number of debt issuers and share of market capitalization excluding the top 10 largest companies

The IMF’s financial development index and its sub-indices may contain measurement errors due to several data limitations (Svirydzenka, 2016). Specifically, the indices rely only on variables with sufficient coverage across countries and years. As a result, some potentially informative indicators are excluded, and some included variables rely on proxies that may not fully capture the intended aspects of financial development. For instance, the financial institution access index uses the number of bank branches and ATMs per 100,000 adults, while other possible measures, such as the number of bank accounts or mobile money usage, are omitted because they do not have sufficient coverage. Similarly, the financial market efficiency index relies on stock market turnover, while bond market measures, such as the bid-ask spread, are not included due to limited data coverage. In addition, the financial institution efficiency index uses basic indicators, including net interest margins, lending-deposit spreads, non-interest income ratios, overhead costs, and profitability measures. These indicators provide only an approximate measure of efficiency, as banks that are technically inefficient may still report profits during favorable economic periods, while efficient banks may incur losses under adverse conditions. Finally, aggregating the sub-indices into a single composite index may amplify these measurement limitations.

in Appendix A. A nonlinear panel data model is first specified to test the hypotheses of a U-shaped relationship between financial development and inequality. The model is specified as follows:

$${GINI}_{i,t}={\theta }_1{FinDev}_{i,t-1}+{\theta }_2{FinDev}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(1)

where GINI denotes the Gini coefficient of disposable income for country i in year t, and FinDev represents financial development. X is a vector of control variables, and α captures country-specific effects. A quadratic term for FinDev is included to account for potential nonlinearities. The FinDev and its quadratic term are lagged by one period to mitigate reverse causality and account for time-lagged effects. It is hardly conceivable that financial development has an immediate effect on incomes.

We employ a fixed-effects estimator, which accounts for unobserved country-specific heterogeneity, to estimate Equation (1). A Hausman test indicates that the fixed-effects model is preferable to the random-effects model. To mitigate reverse causality, FD, its sub-indices, and their squared terms are introduced into the model with a one-period lag. Although this lag structure helps address endogeneity, omitted variable bias and measurement error may persist.

To capture the time persistence in inequality and further address endogeneity, we apply the system GMM estimator proposed by Blundell and Bond (1998), which estimates the model in differences and levels. Variables considered potentially endogenous are FD, its sub-indices, their squared terms, and lagged Gini. The dynamic model in levels is specified as follows:

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FinDev}_{i,t-1}+{\theta }_2{FinDev}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(2)

In the system GMM framework, lagged levels of endogenous variables serve as instruments for the differenced equation, while lagged differences are used as instruments for the level equation. To avoid overfitting due to instrument proliferation, we follow Roodman (2009) and restrict the lag length of the instruments.

Income inequality is measured using the Gini coefficient of disposable income. This widely recognized indicator captures the entire income distribution regardless of a country’s population size or income level.2 The Gini data are drawn from the Standardized World Income Inequality Database (SWIID) due to its high comparability and broad coverage across countries and over time. SWIID employs a Bayesian method to standardize data from multiple sources, using the Luxembourg Income Study as the benchmark (Solt, 2020).

Control variables include investment as a share of GDP, government consumption as a percentage of GDP, trade openness (measured as the ratio of exports and imports to GDP), inflation, inward foreign direct investment (FDI), the human capital index, and the logarithm of real GDP per capita.

Human capital is controlled for to capture the supply-side characteristics of the labor market. While human capital generally reduces inequality through broader educational access (Checchi, 2001; Afonso et al., 2010; Jun et al., 2011; Lustig et al., 2013), it can increase inequality if the gains from education disproportionately favor the already advantaged or highly skilled groups (Becker & Chiswick, 1966; Gregorio & Lee, 2002; Jaumotte et al., 2013; Autor, 2014; Lee & Lee, 2018).

Inflation is another factor that may influence income inequality. It tends to increase inequality when low-income households hold more cash and have wages or transfers less protected against price changes, while richer households hold more diversified assets (Pessino, 1993; Xu & Zou, 2000; Erosa & Ventura, 2002). Conversely, inflation can reduce inequality if inflation is driven by rising input costs outpacing profits (Gramlich, 1974; Blinder & Esaki, 1978).

Real GDP per capita is included, as income inequality depends on the level of economic development (Figini & Görg, 2011). The investment share of GDP is also controlled for, given evidence that investment affects income inequality (Purba et al., 2019; Blotevogel et al., 2020; Josifidis et al., 2021; Shao, 2021). Government consumption is another potential factor influencing income inequality; for example, government expenditure on health and education may help narrow income disparities by improving the overall distribution of human capital (Anderson et al., 2017).

Trade openness is included, given its theoretical relevance to inequality. The Heckscher–Ohlin model provides a framework for investigating how trade openness affects income inequality. This model demonstrates that countries produce export products using relatively cheap and abundant factors of production (Ohlin, 1935). Trade openness influences income inequality due to differences in factors of production and productivity between countries. According to the Stolper-Samuelson theorem, which is derived from the Heckscher-Ohlin model, in countries where skilled workers are intensively employed, openness is expected to increase income inequality by raising skilled wages; conversely, in countries where unskilled workers are relatively abundant, trade openness is anticipated to reduce income inequality by increasing unskilled wages (Stolper & Samuelson, 1941).

FDI is further controlled for due to its potential distributional implications. Feenstra and Hanson (1997) argue that FDI increases the relative demand for skilled workers, thereby contributing to greater income inequality. Figini and Görg (2011) propose that foreign multinational companies introducing new technologies to a host country may increase income inequality, while domestic firms’ adaptation and replication of these technologies in response to FDI inflows may reduce income inequality.

To account for the potential effects of systemic banking crises, we include a dummy variable assigned a value of one for country-years experiencing systemic banking crises. Data on real GDP per capita, investment, government consumption, trade openness, and human capital are obtained from the Penn World Table, given its high cross-country and temporal comparability. Data on inflation and FDI are sourced from the World Bank. Descriptive statistics for all variables are presented in

Table A2. Descriptive statistics.

	Obs.	Mean	Std. Dev.	Min	Max
FD	3852	0.3375	0.2308	0.0000	1.0000
FI	3852	0.4188	0.2318	0.0000	1.0000
FM	3852	0.2428	0.2516	0.0000	1.0000
FIA	3852	0.3337	0.2854	0.0000	1.0000
FIE	3852	0.5514	0.1320	0.0000	0.9254
FID	3852	0.2904	0.2655	0.0000	1.0000
FMA	3852	0.2355	0.2531	0.0000	1.0000
FME	3852	0.2704	0.3415	0.0000	1.0000
FMD	3852	0.2172	0.2624	0.0000	0.9983
Gini	3852	0.3867	0.0854	0.1910	0.6510
Real GDP per capita, log	3852	9.1102	1.1751	6.0593	11.6346
Openness	3852	0.5673	0.5689	0.0214	6.2138
Inflation	3852	0.1837	1.6882	−0.1686	74.8166
Investment share	3852	0.2199	0.0836	0.0045	0.6880
Government consumption share	3852	0.1773	0.0760	0.0052	1.7917
FDI	3852	0.0400	0.1160	−0.4026	2.7936
Human capital	3852	2.4503	0.7098	1.0454	4.3516

in Appendix A.

4. Results

Table 1. Effect of financial development on income inequality: fixed effects estimates.

	Financial Development Index			Financial Institutions Index			Financial Markets Index
	(FD)			(FI)			(FM)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Financial development (t − 1)	0.0365 **	−0.0677 **	−0.0692 **	−0.0160	−0.1376 ***	−0.1425 ***	0.0403 ***	0.0166	0.0168
	(0.0145)	(0.0331)	(0.0341)	(0.0158)	(0.0345)	(0.0345)	(0.0094)	(0.0260)	(0.0266)
Financial development squared (t − 1)		0.1004 ***	0.1022 ***		0.1237 ***	0.1279 ***		0.0262	0.0263
		(0.0297)	(0.0304)		(0.0321)	(0.0318)		(0.0255)	(0.0262)
Financial development (t − 1) × SBCD (t)			0.0108			0.0426			−0.0034
			(0.0351)			(0.0458)			(0.0242)
Financial development squared (t − 1) × SBCD (t)			−0.0143			−0.0402			0.0009
			(0.0338)			(0.0421)			(0.0258)
GDP per capita, log	−0.0047	−0.0032	−0.0032	−0.0007	0.0015	0.0016	−0.0043	−0.0044	−0.0044
	(0.0055)	(0.0054)	(0.0054)	(0.0054)	(0.0053)	(0.0053)	(0.0054)	(0.0053)	(0.0053)
Openness	0.0107 *	0.0098 *	0.0098 *	0.0147 **	0.0127 **	0.0127 **	0.0089	0.0089 *	0.0089 *
	(0.0055)	(0.0053)	(0.0053)	(0.0058)	(0.0056)	(0.0055)	(0.0054)	(0.0053)	(0.0053)
Inflation	0.0006 *	0.0005	0.0005	0.0006 *	0.0005	0.0005	0.0006	0.0005	0.0005
	(0.0004)	(0.0004)	(0.0003)	(0.0003)	(0.0004)	(0.0004)	(0.0004)	(0.0004)	(0.0004)
Investment	0.0043	0.0158	0.0162	−0.0015	0.0116	0.0119	0.0125	0.0146	0.0147
	(0.0175)	(0.0166)	(0.0166)	(0.0180)	(0.0174)	(0.0174)	(0.0171)	(0.0168)	(0.0168)
Government consumption	−0.0285 *	−0.0288 *	−0.0287 *	−0.0212	−0.0211	−0.0209	−0.0299 **	−0.0301 **	−0.0302 **
	(0.0159)	(0.0148)	(0.0148)	(0.0157)	(0.0144)	(0.0143)	(0.0151)	(0.0150)	(0.0151)
FDI	0.0059 *	0.0048 *	0.0047 *	0.0059 *	0.0046 *	0.0042	0.0043	0.0045	0.0045
	(0.0031)	(0.0026)	(0.0026)	(0.0033)	(0.0027)	(0.0028)	(0.0030)	(0.0029)	(0.0029)
Human capital	−0.0153 *	−0.0140	−0.0139	−0.0076	−0.0091	−0.0090	−0.0178 *	−0.0170 *	−0.0170 *
	(0.0092)	(0.0087)	(0.0087)	(0.0091)	(0.0086)	(0.0086)	(0.0090)	(0.0089)	(0.0089)
SBCD	0.0007	0.0003	−0.0009	0.0019	0.0010	−0.0076	0.0002	0.0001	0.0010
	(0.0022)	(0.0022)	(0.0068)	(0.0022)	(0.0022)	(0.0091)	(0.0022)	(0.0021)	(0.0045)
Constant	0.4522 ***	0.4524 ***	0.4523 ***	0.4131 ***	0.4179 ***	0.4182 ***	0.4566 ***	0.4582 ***	0.4580 ***
	(0.0409)	(0.0405)	(0.0405)	(0.0398)	(0.0401)	(0.0401)	(0.0401)	(0.0399)	(0.0399)
Observations	3852	3852	3852	3852	3852	3852	3852	3852	3852
Countries	130	130	130	130	130	130	130	130	130
R-squared	0.0602	0.0879	0.0881	0.0445	0.0828	0.0837	0.0905	0.0930	0.0931
F-test: p-value		0.0009	0.0041		0.0002	0.0003		0.3061	0.5626
Threshold level of financial development (dGINI/dFinDev = 0)		0.34	0.34		0.56	0.56
Percentage of observations above the threshold		41%	41%		28%	28%

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors clustered at the country level are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

,

Table 2. Effect of the dimensions of financial institution development on income inequality: fixed effects estimates.

	Financial Institutions Depth Index			Financial Institutions Efficiency Index			Financial Institutions Access Index
	(FID)			(FIE)			(FIA)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Financial institution development (t − 1)	0.0171	−0.0906 ***	−0.0967 ***	−0.0060	−0.0324	−0.0305	−0.0262 **	−0.1102 ***	−0.1118 ***
	(0.0174)	(0.0313)	(0.0321)	(0.0100)	(0.0417)	(0.0430)	(0.0120)	(0.0325)	(0.0328)
Financial institution development squared (t − 1)		0.1039 ***	0.1104 ***		0.0266	0.0268		0.0854 ***	0.0867 ***
		(0.0271)	(0.0277)		(0.0417)	(0.0447)		(0.0290)	(0.0294)
Financial institution development (t − 1) × SBCD (t)			0.0502			−0.0086			0.0241
			(0.0317)			(0.0682)			(0.0319)
Financial institution development squared (t − 1) × SBCD (t)			−0.0610 *			−0.0059			−0.0214
			(0.0338)			(0.0613)			(0.0300)
GDP per capita, log	−0.0027	−0.0009	−0.0008	−0.0017	−0.0019	−0.0019	0.0012	0.0041	0.0041
	(0.0053)	(0.0051)	(0.0051)	(0.0055)	(0.0054)	(0.0054)	(0.0053)	(0.0052)	(0.0052)
Openness	0.0130 **	0.0124 **	0.0123 **	0.0139 **	0.0141 **	0.0142 **	0.0154 ***	0.0140 **	0.0141 **
	(0.0058)	(0.0055)	(0.0055)	(0.0057)	(0.0057)	(0.0058)	(0.0058)	(0.0055)	(0.0056)
Inflation	0.0006 *	0.0005	0.0005	0.0006 *	0.0006 *	0.0005 *	0.0005 *	0.0005	0.0005
	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0003)	(0.0003)
Investment	−0.0000	0.0120	0.0127	−0.0010	−0.0008	−0.0003	−0.0001	0.0066	0.0068
	(0.0178)	(0.0169)	(0.0168)	(0.0183)	(0.0182)	(0.0183)	(0.0177)	(0.0174)	(0.0174)
Government consumption	−0.0248	−0.0269 *	−0.0272 *	−0.0223	−0.0232	−0.0236	−0.0207	−0.0198	−0.0198
	(0.0158)	(0.0151)	(0.0150)	(0.0152)	(0.0156)	(0.0157)	(0.0160)	(0.0161)	(0.0160)
FDI	0.0064 *	0.0051 *	0.0044	0.0061 *	0.0061 *	0.0060 *	0.0053	0.0043	0.0039
	(0.0032)	(0.0027)	(0.0029)	(0.0033)	(0.0033)	(0.0033)	(0.0033)	(0.0031)	(0.0032)
Human capital	−0.0122	−0.0110	−0.0109	−0.0094	−0.0088	−0.0090	−0.0069	−0.0057	−0.0057
	(0.0097)	(0.0086)	(0.0085)	(0.0091)	(0.0089)	(0.0090)	(0.0090)	(0.0091)	(0.0091)
SBCD	0.0012	0.0014	−0.0041	0.0015	0.0014	0.0077	0.0021	0.0012	−0.0028
	(0.0022)	(0.0022)	(0.0046)	(0.0021)	(0.0021)	(0.0188)	(0.0022)	(0.0023)	(0.0043)
Constant	0.4332 ***	0.4269 ***	0.4266 ***	0.4244 ***	0.4303 ***	0.4298 ***	0.3953 ***	0.3769 ***	0.3775 ***
	(0.0394)	(0.0384)	(0.0382)	(0.0395)	(0.0395)	(0.0397)	(0.0398)	(0.0399)	(0.0399)
Observations	3852	3852	3852	3852	3852	3852	3852	3852	3852
Countries	130	130	130	130	130	130	130	130	130
R-squared	0.0447	0.0841	0.0873	0.0421	0.0431	0.0442	0.0569	0.0871	0.0878
F-test: p-value		0.0002	0.0005		0.5251	0.8208		0.0038	0.0130
Threshold level for dimensions of financial development		0.44	0.44					0.65	0.64
Percentage of observations above the threshold		26%	26%					19%	19%

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors clustered at the country level are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

and

Table 3. Effect of the dimensions of financial market development on income inequality: fixed effects estimates.

	Financial Markets Depth Index			Financial Markets Efficiency Index			Financial Markets Access Index
	(FMD)			(FME)			(FMA)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Financial market development (t − 1)	0.0391 ***	0.0370 *	0.0371 *	0.0145 **	0.0146	0.0131	0.0325 ***	0.0905 ***	0.0895 ***
	(0.0079)	(0.0207)	(0.0213)	(0.0062)	(0.0200)	(0.0208)	(0.0097)	(0.0264)	(0.0267)
Financial market development squared (t − 1)		0.0023	0.0024		−0.0001	0.0016		−0.0644 **	−0.0631 **
		(0.0199)	(0.0208)		(0.0168)	(0.0177)		(0.0259)	(0.0263)
Financial market development (t − 1) × SBCD (t)			−0.0027			0.0139			0.0122
			(0.0202)			(0.0220)			(0.0233)
Financial market development squared (t − 1) × SBCD (t)			0.0005			−0.0152			−0.0157
			(0.0205)			(0.0214)			(0.0238)
GDP per capita, log	−0.0055	−0.0055	−0.0054	−0.0020	−0.0020	−0.0021	−0.0043	−0.0036	−0.0036
	(0.0053)	(0.0053)	(0.0053)	(0.0053)	(0.0053)	(0.0053)	(0.0054)	(0.0053)	(0.0053)
Openness	0.0079	0.0079	0.0080	0.0120 **	0.0120 **	0.0120 **	0.0115 **	0.0108 *	0.0108 *
	(0.0052)	(0.0051)	(0.0051)	(0.0058)	(0.0058)	(0.0058)	(0.0053)	(0.0055)	(0.0055)
Inflation	0.0006	0.0005	0.0005	0.0006	0.0006	0.0006	0.0006 *	0.0006 *	0.0006 *
	(0.0004)	(0.0004)	(0.0003)	(0.0004)	(0.0004)	(0.0004)	(0.0004)	(0.0004)	(0.0004)
Investment	0.0137	0.0139	0.0139	0.0031	0.0031	0.0031	0.0080	0.0055	0.0058
	(0.0167)	(0.0166)	(0.0165)	(0.0178)	(0.0176)	(0.0176)	(0.0175)	(0.0177)	(0.0177)
Government consumption	−0.0320 **	−0.0319 **	−0.0319 **	−0.0248	−0.0248	−0.0247	−0.0279 *	−0.0297 *	−0.0296 *
	(0.0153)	(0.0153)	(0.0153)	(0.0153)	(0.0153)	(0.0153)	(0.0153)	(0.0155)	(0.0155)
FDI	0.0036	0.0036	0.0036	0.0065 **	0.0065 **	0.0063 **	0.0035	0.0057 *	0.0054 *
	(0.0029)	(0.0030)	(0.0029)	(0.0030)	(0.0030)	(0.0030)	(0.0034)	(0.0032)	(0.0031)
Human capital	−0.0187 **	−0.0187 **	−0.0187 **	−0.0122	−0.0122	−0.0121	−0.0150	−0.0183 *	−0.0183 *
	(0.0087)	(0.0086)	(0.0086)	(0.0091)	(0.0091)	(0.0091)	(0.0093)	(0.0098)	(0.0098)
SBCD	−0.0001	−0.0001	0.0005	0.0009	0.0009	0.0001	0.0012	0.0010	−0.0001
	(0.0021)	(0.0021)	(0.0037)	(0.0022)	(0.0022)	(0.0035)	(0.0021)	(0.0021)	(0.0042)
Constant	0.4725 ***	0.4721 ***	0.4719 ***	0.4276 ***	0.4276 ***	0.4280 ***	0.4511 ***	0.4483 ***	0.4483 ***
	(0.0406)	(0.0406)	(0.0406)	(0.0390)	(0.0384)	(0.0384)	(0.0406)	(0.0403)	(0.0404)
Observations	3852	3852	3852	3852	3852	3852	3852	3852	3852
Countries	130	130	130	130	130	130	130	130	130
R-squared	0.1018	0.1018	0.1019	0.0573	0.0573	0.0577	0.0701	0.0856	0.0859
F-test: p-value		0.9102	0.9900		0.9943	0.7622		0.0139	0.0185
Threshold level for dimensions of financial development								0.70	0.71
Percentage of observations above the threshold								6%	5%

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors clustered at the country level are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

show the fixed-effects estimation results. Table 1 presents the results from estimating Equation (1). Column 2 reveals a U-shaped relationship: financial development initially reduces inequality, but its marginal effect becomes positive once FD exceeds a threshold of 0.34. This implies that while moderate financial development promotes more equitable income distribution, beyond a certain level it may disproportionately benefit higher-income groups, widening inequality. These results provide evidence of a nonlinear effect, addressing our research question. Column 3 adds interaction terms between financial development and a banking crisis dummy; the main effects remain significant in tranquil periods, suggesting that systemic banking crises do not drive the U-shaped relationship.

Similarly, column 5 indicates a U-shaped relationship between financial institution development and inequality, with the marginal effect turning positive once FI exceeds the threshold of 0.56. This indicates that, in the early stages, strengthening financial institutions helps reduce inequality; however, beyond a certain point, the benefits disproportionately favor higher-income groups, leading to an increase in inequality and supporting the nonlinear relationship. Column 6 shows that the main effects remain significant in tranquil periods after including the interaction terms. By contrast, columns 7–9 suggest that financial market development has no robust effect on inequality.

These findings suggest that the nonlinear effect of financial development is mainly driven by financial institutions rather than financial markets, offering a more nuanced understanding of how financial development affects income inequality.

Next, we assess how key dimensions of financial institution development—depth, efficiency, and access—affect inequality by incorporating the three sub-indices of FI into the model, as specified in Equations (3)–(5):

$${GINI}_{i,t}={\theta }_1{FID}_{i,t-1}+{\theta }_2{FID}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(3)

$${GINI}_{i,t}={\theta }_1{FIE}_{i,t-1}+{\theta }_2{FIE}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(4)

$${GINI}_{i,t}={\theta }_1{FIA}_{i,t-1}+{\theta }_2{FIA}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(5)

Table 2 presents the results from estimating Equations (3)–(5). Columns 2 and 8 indicate that the marginal effects of financial institution depth and access become positive when FID exceeds 0.44 and FIA exceeds 0.65, respectively. This implies that improving financial institution depth and access initially helps reduce inequality, but beyond these respective thresholds, further development exacerbates inequality. These results address our research question and support our two hypotheses that financial depth and financial access have U-shaped relationships with income inequality. In columns 3 and 9, the main effects remain significant in tranquil periods after including interaction terms, with the threshold for FIA decreasing slightly to 0.64, which indicates that these U-shaped relationships are not driven by systemic banking crises. The efficiency of financial institutions, however, shows no significant effect.

These findings suggest that the nonlinear relationship between financial institution development and inequality is mainly driven by the depth and access dimensions of financial institutions rather than the efficiency dimension.

Furthermore, to explore the effects of various dimensions of financial market development—depth, efficiency, and access—on inequality, we extend the model by incorporating the three sub-indices of FM, as presented below:

$${GINI}_{i,t}={\theta }_1{FMD}_{i,t-1}+{\theta }_2{FMD}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(6)

$${GINI}_{i,t}={\theta }_1{FME}_{i,t-1}+{\theta }_2{FME}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(7)

$${GINI}_{i,t}={\theta }_1{FMA}_{i,t-1}+{\theta }_2{FMA}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(8)

Table 3 presents the results from estimating Equations (6)–(8). Financial market efficiency has no robust effect on inequality. In contrast, financial market access exhibits an inverted U-shaped pattern, contrary to our second hypothesis. This pattern suggests that at early stages, greater financial market access increases inequality, likely because it primarily benefits higher-income groups. Beyond the threshold of 0.70, access expands to a broader population, reducing inequality. However, this pattern is not robust: our system GMM results indicate that financial market access reduces inequality, with no evidence of a nonlinear effect. This discrepancy may arise because the system GMM approach addresses endogeneity more effectively and can capture the time persistence in the Gini coefficient, which the fixed-effects model cannot.

Financial market depth, however, tends to increase inequality, with no evidence of a nonlinear effect, contrary to our first hypothesis, indicating that it primarily advantages higher-income groups. Again, this result is not robust: the system GMM estimates show a U-shaped effect of financial market depth on inequality, although it is not statistically significant. The same explanation regarding endogeneity and time persistence applies.

Table 4. Effect of financial development on income inequality: system GMM estimates.

	Financial Development Index			Financial Institutions Index			Financial Markets Index
	(FD)			(FI)			(FM)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Gini (t − 1)	0.7728 ***	0.7169 ***	0.7411 ***	0.7690 ***	0.7438 ***	0.7886 ***	0.7169 ***	0.7795 ***	0.8199 ***
	(0.0605)	(0.0903)	(0.0804)	(0.0712)	(0.0732)	(0.0532)	(0.0866)	(0.0647)	(0.0486)
Financial development (t − 1)	0.0120	−0.1929 **	−0.2045 **	−0.0551	−0.1540 ***	−0.1720 ***	−0.0008	−0.0217	−0.0111
	(0.0122)	(0.0821)	(0.0834)	(0.0483)	(0.0520)	(0.0507)	(0.0092)	(0.0346)	(0.0257)
Financial development squared (t − 1)		0.1599 **	0.1712 **		0.1239 ***	0.1393 ***		0.0201	0.0101
		(0.0666)	(0.0698)		(0.0414)	(0.0420)		(0.0341)	(0.0259)
Financial development (t − 1) × SBCD (t)			0.1733			0.2903 *			0.0422
			(0.1306)			(0.1558)			(0.0285)
Financial development squared (t − 1) × SBCD (t)			−0.1643			−0.2567 *			−0.0388
			(0.1215)			(0.1380)			(0.0268)
GDP per capita, log	−0.0186 ***	−0.0139 **	−0.0112 *	−0.0113 **	−0.0155 **	−0.0115 **	−0.0223 ***	−0.0171 ***	−0.0148 ***
	(0.0061)	(0.0069)	(0.0060)	(0.0056)	(0.0063)	(0.0049)	(0.0082)	(0.0059)	(0.0049)
Openness	−0.0007	−0.0008	−0.0008	0.0008	−0.0000	0.0003	−0.0006	−0.0007	−0.0006
	(0.0031)	(0.0042)	(0.0040)	(0.0030)	(0.0037)	(0.0032)	(0.0039)	(0.0033)	(0.0028)
Inflation	0.0007	0.0009	0.0008	0.0006	0.0006	0.0005	0.0009	0.0007	0.0006
	(0.0005)	(0.0006)	(0.0005)	(0.0005)	(0.0006)	(0.0005)	(0.0006)	(0.0005)	(0.0004)
Investment	−0.0035	0.0246	0.0238	0.0111	0.0131	0.0120	0.0002	0.0049	0.0032
	(0.0176)	(0.0254)	(0.0236)	(0.0204)	(0.0222)	(0.0188)	(0.0214)	(0.0174)	(0.0146)
Government consumption	−0.0546 *	−0.0752 *	−0.0668 **	−0.0702 *	−0.0570 *	−0.0451 *	−0.0765 *	−0.0611 *	−0.0481 **
	(0.0282)	(0.0385)	(0.0336)	(0.0361)	(0.0301)	(0.0232)	(0.0393)	(0.0317)	(0.0244)
FDI	−0.0005	−0.0022	−0.0044	−0.0009	−0.0022	−0.0049	−0.0007	−0.0003	−0.0011
	(0.0031)	(0.0054)	(0.0053)	(0.0036)	(0.0048)	(0.0043)	(0.0040)	(0.0033)	(0.0028)
Human capital	0.0176 *	0.0260 **	0.0227 **	0.0229 **	0.0241 **	0.0195 **	0.0245 **	0.0205 **	0.0176 **
	(0.0095)	(0.0125)	(0.0110)	(0.0115)	(0.0103)	(0.0081)	(0.0117)	(0.0091)	(0.0078)
SBCD	−0.0025	−0.0022	−0.0344	−0.0019	−0.0033	−0.0658 *	−0.0029	−0.0018	−0.0079
	(0.0024)	(0.0032)	(0.0274)	(0.0025)	(0.0030)	(0.0358)	(0.0031)	(0.0023)	(0.0061)
Constant	0.2213 ***	0.2186 ***	0.1937 ***	0.1684 ***	0.2249 ***	0.1836 ***	0.2662 ***	0.2034 ***	0.1706 ***
	(0.0592)	(0.0715)	(0.0625)	(0.0514)	(0.0667)	(0.0492)	(0.0822)	(0.0565)	(0.0452)
AR (2): p-value	0.478	0.715	0.587	0.895	0.603	0.181	0.881	0.542	0.210
Hansen test: p-value	0.010	0.484	0.795	0.042	0.600	0.835	0.054	0.045	0.062
Difference-in-Hansen test (for levels, excluding group): p-value	0.907	0.843	0.935	0.610	0.939	0.983	0.433	0.192	0.272
Difference-in-Hansen test (for levels, difference): p-value	0.001	0.142	0.313	0.008	0.161	0.260	0.017	0.038	0.037
Difference-in-Hansen test (iv, excluding group): p-value		0.323	0.423		0.699	0.613		0.557	0.102
Difference-in-Hansen test (iv, difference): p-value		0.481	0.928		0.509	0.830		0.030	0.142
Instruments	16	20	28	16	20	28	16	20	28
Observations	3803	3803	3803	3803	3803	3803	3803	3803	3803
Countries	130	130	130	130	130	130	130	130	130
Threshold level of financial development (dGINI/dFinDev = 0)		0.60	0.60		0.62	0.62
Percentage of observations above the threshold		16%	16%		23%	23%

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

,

Table 5. Effect of the dimensions of financial institution development on income inequality: system GMM estimates.

	Financial Institutions Depth Index			Financial Institutions Efficiency Index			Financial Institutions Access Index
	(FID)			(FIE)			(FIA)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Gini (t − 1)	0.7716 ***	0.6613 ***	0.7276 ***	0.6640 ***	0.7642 ***	0.8385 ***	0.7480 ***	0.7689 ***	0.8001 ***
	(0.0648)	(0.1117)	(0.0759)	(0.1117)	(0.0718)	(0.0379)	(0.0709)	(0.0646)	(0.0525)
Financial institution development (t − 1)	0.0207	−0.1457 **	−0.1213 **	−0.0566 ***	−0.1979 ***	−0.1862 ***	−0.0248	−0.0762 ***	−0.0850 ***
	(0.0148)	(0.0674)	(0.0488)	(0.0207)	(0.0669)	(0.0598)	(0.0180)	(0.0286)	(0.0279)
Financial institution development squared (t − 1)		0.1214 **	0.1059 **		0.1749 ***	0.1725 ***		0.0553 **	0.0612 ***
		(0.0548)	(0.0418)		(0.0594)	(0.0551)		(0.0227)	(0.0217)
Financial institution development (t − 1) × SBCD (t)			0.1113 *			0.3534			0.0934 *
			(0.0585)			(0.2669)			(0.0521)
Financial institution development squared (t − 1) × SBCD (t)			−0.1077 *			−0.3199			−0.0781 *
			(0.0551)			(0.2399)			(0.0453)
GDP per capita, log	−0.0201 ***	−0.0148 **	−0.0112 **	−0.0254 ***	−0.0219 ***	−0.0139 ***	−0.0160 **	−0.0147 **	−0.0110 **
	(0.0064)	(0.0068)	(0.0051)	(0.0099)	(0.0072)	(0.0048)	(0.0068)	(0.0061)	(0.0049)
Openness	−0.0026	0.0013	0.0005	−0.0006	−0.0004	−0.0005	−0.0019	−0.0012	−0.0011
	(0.0039)	(0.0043)	(0.0034)	(0.0049)	(0.0035)	(0.0023)	(0.0033)	(0.0032)	(0.0027)
Inflation	0.0007 *	0.0010	0.0008	0.0006	0.0005	0.0005	0.0007	0.0007	0.0006
	(0.0004)	(0.0007)	(0.0005)	(0.0005)	(0.0004)	(0.0004)	(0.0005)	(0.0006)	(0.0005)
Investment	−0.0017	0.0070	0.0035	0.0193	0.0083	0.0012	0.0049	0.0053	0.0052
	(0.0178)	(0.0269)	(0.0213)	(0.0289)	(0.0208)	(0.0141)	(0.0195)	(0.0181)	(0.0155)
Government consumption	−0.0506 *	−0.0925 **	−0.0698 **	−0.0834 *	−0.0710 **	−0.0457 **	−0.0679 **	−0.0578 **	−0.0481 **
	(0.0282)	(0.0469)	(0.0329)	(0.0447)	(0.0328)	(0.0186)	(0.0312)	(0.0286)	(0.0231)
FDI	0.0004	−0.0035	−0.0045	−0.0036	−0.0000	−0.0005	0.0010	−0.0006	−0.0022
	(0.0032)	(0.0056)	(0.0045)	(0.0052)	(0.0039)	(0.0026)	(0.0035)	(0.0039)	(0.0035)
Human capital	0.0180 *	0.0216 *	0.0148 *	0.0292 **	0.0279 ***	0.0169 **	0.0237 **	0.0252 **	0.0211 **
	(0.0093)	(0.0115)	(0.0084)	(0.0140)	(0.0106)	(0.0075)	(0.0106)	(0.0104)	(0.0085)
SBCD	−0.0024	−0.0030	−0.0190 *	−0.0058	−0.0041	−0.0922	−0.0018	−0.0025	−0.0189 *
	(0.0024)	(0.0037)	(0.0114)	(0.0042)	(0.0029)	(0.0699)	(0.0026)	(0.0025)	(0.0111)
Constant	0.2324 ***	0.2507 ***	0.2009 ***	0.3326 ***	0.2866 ***	0.2033 ***	0.2054 ***	0.1855 ***	0.1505 ***
	(0.0627)	(0.0843)	(0.0591)	(0.1087)	(0.0823)	(0.0495)	(0.0647)	(0.0576)	(0.0454)
AR (2): p-value	0.477	0.916	0.154	0.382	0.669	0.177	0.897	0.957	0.311
Hansen test: p-value	0.019	0.771	0.708	0.655	0.372	0.419	0.053	0.097	0.294
Difference-in-Hansen test (for levels, excluding group): p-value	0.486	0.891	0.958	0.972	0.969	0.987	0.537	0.638	0.893
Difference-in-Hansen test (for levels, difference): p-value	0.004	0.363	0.181	0.216	0.052	0.032	0.012	0.018	0.034
Difference-in-Hansen test (iv, excluding group): p-value		0.781	0.424		0.684	0.793		0.088	0.123
Difference-in-Hansen test (iv, difference): p-value		0.685	0.817		0.291	0.160		0.160	0.666
Instruments	16	20	28	16	20	28	16	20	28
Observations	3803	3803	3803	3803	3803	3803	3803	3803	3803
Countries	130	130	130	130	130	130	130	130	130
Threshold level for dimensions of financial development		0.60	0.57		0.57	0.52		0.69	0.69
Percentage of observations above the threshold		17%	20%		48%	64%		16%	16%

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

and

Table 6. Effect of the dimensions of financial market development on income inequality: system GMM estimates.

	Financial Markets Depth Index			Financial Markets Efficiency Index			Financial Markets Access Index
	(FMD)			(FME)			(FMA)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Gini (t − 1)	0.6730 ***	0.6936 ***	0.7012 ***	0.7478 ***	0.7500 ***	0.7904 ***	0.7009 ***	0.7733 ***	0.8234 ***
	(0.1074)	(0.1001)	(0.0989)	(0.0806)	(0.0849)	(0.0644)	(0.0940)	(0.0615)	(0.0439)
Financial market development (t − 1)	−0.0011	−0.0232	−0.0296	0.0022	−0.0105	−0.0061	−0.0221 *	−0.0417 *	−0.0249
	(0.0083)	(0.0229)	(0.0239)	(0.0075)	(0.0251)	(0.0199)	(0.0125)	(0.0235)	(0.0188)
Financial market development squared (t − 1)		0.0232	0.0293		0.0133	0.0107		0.0312	0.0169
		(0.0229)	(0.0243)		(0.0178)	(0.0147)		(0.0206)	(0.0167)
Financial market development (t − 1) × SBCD (t)			0.0653			0.0365 *			0.0270
			(0.0442)			(0.0219)			(0.0249)
Financial market development squared (t − 1) × SBCD (t)			−0.0598			−0.0315 *			−0.0248
			(0.0402)			(0.0183)			(0.0228)
GDP per capita, log	−0.0246 ***	−0.0222 ***	−0.0213 **	−0.0202 ***	−0.0197 ***	−0.0167 ***	−0.0209 ***	−0.0162 ***	−0.0142 ***
	(0.0094)	(0.0085)	(0.0083)	(0.0071)	(0.0069)	(0.0056)	(0.0081)	(0.0059)	(0.0047)
Openness	−0.0006	−0.0006	−0.0007	−0.0005	−0.0004	−0.0002	−0.0003	−0.0007	−0.0005
	(0.0045)	(0.0042)	(0.0042)	(0.0035)	(0.0035)	(0.0029)	(0.0043)	(0.0036)	(0.0029)
Inflation	0.0010	0.0009	0.0009	0.0008	0.0008	0.0006	0.0010 *	0.0008 *	0.0007 *
	(0.0006)	(0.0006)	(0.0006)	(0.0005)	(0.0005)	(0.0004)	(0.0006)	(0.0004)	(0.0004)
Investment	−0.0016	0.0012	0.0011	−0.0009	−0.0008	−0.0019	0.0041	0.0080	0.0067
	(0.0248)	(0.0232)	(0.0225)	(0.0193)	(0.0193)	(0.0164)	(0.0230)	(0.0183)	(0.0148)
Government consumption	−0.0875 **	−0.0820 *	−0.0795 *	−0.0663 *	−0.0659 *	−0.0526 *	−0.0922 **	−0.0693 **	−0.0531 **
	(0.0446)	(0.0419)	(0.0411)	(0.0357)	(0.0371)	(0.0286)	(0.0439)	(0.0308)	(0.0234)
FDI	−0.0007	−0.0002	−0.0016	−0.0003	−0.0007	0.0001	0.0036	−0.0011	−0.0010
	(0.0047)	(0.0044)	(0.0047)	(0.0040)	(0.0046)	(0.0036)	(0.0043)	(0.0039)	(0.0032)
Human capital	0.0261 *	0.0241 *	0.0231 *	0.0217 **	0.0215 **	0.0178 **	0.0259 **	0.0215 **	0.0191 **
	(0.0134)	(0.0123)	(0.0121)	(0.0101)	(0.0100)	(0.0082)	(0.0124)	(0.0095)	(0.0077)
SBCD	−0.0035	−0.0032	−0.0112	−0.0027	−0.0027	−0.0069	−0.0025	−0.0014	−0.0054
	(0.0036)	(0.0034)	(0.0089)	(0.0027)	(0.0027)	(0.0050)	(0.0031)	(0.0023)	(0.0055)
Constant	0.3033 ***	0.2789 ***	0.2706 ***	0.2396 ***	0.2355 ***	0.1998 ***	0.2633 ***	0.1995 ***	0.1630 ***
	(0.0996)	(0.0909)	(0.0903)	(0.0730)	(0.0727)	(0.0564)	(0.0822)	(0.0544)	(0.0419)
AR (2): p-value	0.990	0.864	0.482	0.716	0.659	0.165	0.746	0.535	0.251
Hansen test: p-value	0.110	0.339	0.624	0.016	0.054	0.017	0.322	0.300	0.354
Difference-in-Hansen test (for levels, excluding group): p-value	0.669	0.634	0.825	0.167	0.275	0.116	0.711	0.791	0.954
Difference-in-Hansen test (for levels, difference): p-value	0.024	0.132	0.246	0.012	0.030	0.021	0.107	0.068	0.033
Difference-in-Hansen test (iv, excluding group): p-value		0.066	0.375		0.379	0.059		0.610	0.309
Difference-in-Hansen test (iv, difference): p-value		0.581	0.752		0.043	0.054		0.235	0.421
Instruments	16	20	28	16	20	28	16	20	28
Observations	3803	3803	3803	3803	3803	3803	3803	3803	3803
Countries	130	130	130	130	130	130	130	130	130
Threshold level for dimensions of financial development
Percentage of observations above the threshold

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

present the results from the dynamic system GMM estimations. Table 4 shows the estimation results for Equation (2). Column 2 indicates that financial development initially alleviates inequality, but its marginal effect becomes positive once FD exceeds a threshold of 0.60. This suggests that financial development at early stages improves income distribution, whereas further expansion disproportionately benefits higher-income groups, intensifying inequality. These results provide evidence of a nonlinear relationship, addressing our research question. Comparing columns 1 and 2 highlights the importance of including the squared term, confirming the presence of nonlinearity. After including interaction terms (column 3), the main effects remain significant in tranquil periods, suggesting that systemic banking crises do not drive the nonlinear effect of financial development.

In column 5, the marginal effect of financial institution development turns positive beyond a threshold of 0.62. This indicates that financial institution development initially reduces inequality, but beyond the threshold, it disproportionately favors higher-income groups, worsening inequality. After including interaction terms (column 6), the main effects remain significant in tranquil periods. In contrast, columns 7–9 provide no evidence that financial market development affects inequality.

These results indicate that the nonlinear effect of financial development is primarily driven by financial institutions rather than financial markets.

To further explore the effects of the dimensions of financial development on inequality, we extend the dynamic model by incorporating the sub-indices of FI and FM as follows:

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FID}_{i,t-1}+{\theta }_2{FID}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(9)

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FIE}_{i,t-1}+{\theta }_2{FIE}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(10)

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FIA}_{i,t-1}+{\theta }_2{FIA}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(11)

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FMD}_{i,t-1}+{\theta }_2{FMD}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(12)

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FME}_{i,t-1}+{\theta }_2{FME}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(13)

$${GINI}_{i,t}=\beta {GINI}_{i,t-1}+{\theta }_1{FMA}_{i,t-1}+{\theta }_2{FMA}_{i,t-1}^2+{\theta }_3{X}_{i,t}+{\alpha }_i+{\epsilon }_{i,t}$$

(14)

Table 5 provides the results from estimating Equations (9)–(11). The marginal effect of financial institution depth becomes positive above a threshold of 0.60 (column 2), which decreases slightly to 0.57 after including interaction terms (column 3), with main effects remaining significant in tranquil periods. This indicates that increasing financial institution depth initially reduces inequality, but beyond the threshold, it worsens inequality. These findings address our research question and provide support for our first hypothesis that financial depth exhibits a U-shaped relationship with income inequality.

Similarly, the marginal effect of financial institution efficiency becomes positive beyond a threshold of 0.57 (column 5) and declines slightly to 0.52 when interaction terms are included (column 6). However, this pattern is not robust, as it differs from the results obtained using fixed-effects estimation. Likewise, the marginal effect of financial institution access becomes positive once it exceeds the threshold of 0.69 (column 8). This suggests that improving financial institution access initially lowers inequality, but once the threshold is surpassed, it intensifies inequality. These results contribute to answering our research question and support our second hypothesis that financial access follows a U-shaped relationship with income inequality. The main effects remain significant in tranquil periods, even after including the interaction terms (column 9).

The findings suggest that the nonlinear relationship between financial institution development and inequality is mainly driven by the depth and access dimensions.

Table 6 presents the results from Equations (12)–(14). There is no robust evidence that financial market depth, financial market efficiency, or financial market access have a statistically significant effect on inequality. These findings indicate that none of these dimensions exhibit a nonlinear effect on inequality and do not support either of our two hypotheses.

5. Discussion

Our findings are consistent with the U-shaped relationship between financial depth and inequality proposed by Tan and Law (2012) and Cihak and Sahay (2020). However, unlike these studies, which focus solely on financial depth, this study examines the effect of overall financial development and its depth, access, and efficiency dimensions on inequality. In addition to confirming a U-shaped relationship between financial depth and inequality, our results indicate that both overall financial development and financial institution access similarly exhibit a U-shaped relationship with inequality.

Our findings are also consistent with Brei et al. (2023), who document a U-shaped relationship between overall financial development and inequality and show that financial efficiency does not exert a robust effect on inequality. In contrast to Brei et al. (2023), our results indicate that financial institution depth exerts a statistically significant effect on inequality.

One possible reason for this difference is that Brei et al. (2023) employ two proxies for financial depth—bank credit to GDP and stock market capitalization to GDP—which may overlook key aspects of financial depth. One possible reason for this difference is that Brei et al. (2023) employ two proxies for financial depth—bank credit to GDP and stock market capitalization to GDP—thereby overlooking the role of non-bank financial institutions and potentially omitting key aspects of financial depth. They also use a composite index encompassing both financial institutions and financial markets, which does not distinguish between the two. In contrast, our study uses the IMF’s Financial Institutions Depth Index and Financial Markets Depth Index separately, providing a more comprehensive measure of financial depth and allowing us to distinguish between financial institutions and financial markets. Another possible reason is that Brei et al. (2023) focus on 97 countries from 1989 to 2012, whereas our study covers a larger set of countries over a longer period. Furthermore, unlike Brei et al. (2023), we find that financial institution access exerts a statistically significant U-shaped effect on inequality, likely because our approach distinguishes between financial institutions and financial markets, while their composite index does not.

We find that overall financial development, financial institution development, financial institution depth, and financial institution access have U-shaped effects on inequality, as measured by the Gini coefficient. These effects are robust across different estimators and are not driven by systemic banking crises. The Gini coefficient is the most widely used measure of income inequality because it captures the entire income distribution, regardless of a country’s income level or population size. Therefore, our main estimation and policy implications are based on the Gini coefficient rather than on other indicators.

In contrast, percentile ratios capture differences only at specific points in the distribution, ignoring the rest of the income spectrum (Trapeznikova, 2019). However, the Gini coefficient may mask shifts within the distribution. To address this concern, we further check for robustness by assessing whether the statistically significant U-shaped effects of overall financial development, financial institution development, financial institution depth, and financial institution access identified with the Gini coefficient also hold for the top 10% income share and the Palma ratio from the World Inequality Database (WID).

The system GMM results in

Table A3. Effect of financial development on top 10% income share and Palma ratio: system GMM estimates.

	Financial Development Index		Financial Institutions Index		Financial Institutions Depth Index		Financial Institutions Access Index
	(FD)		(FI)		(FID)		(FIA)
	Top10%	Palma	Top10%	Palma	Top10%	Palma	Top10%	Palma
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Gini (t − 1)	0.9919 ***	0.9816 ***	0.9434 ***	1.0039 ***	0.9449 ***	0.9966 ***	0.9315 ***	0.9804 ***
	(0.0179)	(0.0261)	(0.0271)	(0.0180)	(0.0268)	(0.0227)	(0.0413)	(0.0191)
Financial development (t − 1)	0.0496	−0.9635	−0.0309	0.3655	0.0036	0.3017	−0.0319	−0.4530
	(0.0338)	(1.6142)	(0.0427)	(0.7351)	(0.0218)	(0.6597)	(0.0347)	(0.5129)
Financial development squared (t − 1)	−0.0355	0.8101	0.0363	−0.4219	0.0074	−0.1451	0.0316	0.4644
	(0.0266)	(1.3520)	(0.0381)	(0.7586)	(0.0197)	(0.6165)	(0.0324)	(0.5012)
Financial development (t − 1) × SBCD (t)	−0.0285	−1.2261 *	0.0197	−1.6567 *	0.0116	−0.6710 *	0.0113	−0.5938
	(0.0230)	(0.6957)	(0.0564)	(1.0063)	(0.0176)	(0.3794)	(0.0229)	(0.4102)
Financial development squared (t − 1) × SBCD (t)	0.0189	1.1043	−0.0273	1.4598	−0.0234	0.6197	−0.0161	0.4551
	(0.0228)	(0.6865)	(0.0530)	(0.9344)	(0.0187)	(0.4012)	(0.0216)	(0.3830)
GDP per capita, log	0.0030	0.2301 *	−0.0007	0.1709 **	−0.0009	0.1314	−0.0011	0.1792 ***
	(0.0029)	(0.1393)	(0.0037)	(0.0684)	(0.0030)	(0.0931)	(0.0054)	(0.0645)
Openness	0.0002	0.0129	−0.0001	0.0099	−0.0011	−0.0073	0.0002	0.0155
	(0.0009)	(0.0252)	(0.0009)	(0.0226)	(0.0010)	(0.0267)	(0.0012)	(0.0233)
Inflation	0.0000	0.0006	0.0000	0.0004	0.0000	0.0005	0.0000	0.0008
	(0.0000)	(0.0010)	(0.0000)	(0.0009)	(0.0000)	(0.0008)	(0.0000)	(0.0010)
Investment	−0.0149 ***	−0.1966	−0.0101 *	−0.1643	−0.0111 **	−0.1964	−0.0111 **	−0.2836 *
	(0.0044)	(0.1767)	(0.0059)	(0.1342)	(0.0045)	(0.1823)	(0.0055)	(0.1566)
Government consumption	0.0175 **	0.3857 **	0.0076	0.4136 **	0.0085	0.4679 ***	0.0032	0.4129 **
	(0.0070)	(0.1789)	(0.0063)	(0.1668)	(0.0066)	(0.1636)	(0.0087)	(0.1609)
FDI	0.0021	0.0216	0.0011	0.0259	0.0017	0.0197	0.0008	0.0071
	(0.0020)	(0.0484)	(0.0018)	(0.0469)	(0.0021)	(0.0445)	(0.0019)	(0.0543)
Human capital	−0.0122 ***	−0.3831 **	−0.0049	−0.3017 ***	−0.0061	−0.3007 *	−0.0020	−0.3426 ***
	(0.0035)	(0.1889)	(0.0062)	(0.1106)	(0.0041)	(0.1791)	(0.0092)	(0.1257)
SBCD	0.0053	0.2313 *	−0.0037	0.3594 *	−0.0014	0.1031 *	−0.0025	0.1066
	(0.0040)	(0.1287)	(0.0117)	(0.2120)	(0.0026)	(0.0605)	(0.0044)	(0.0764)
Constant	−0.0052	−0.9587 **	0.0469	−0.9682 *	0.0450	−0.5774	0.0495	−0.6984 *
	(0.0221)	(0.4888)	(0.0373)	(0.5119)	(0.0294)	(0.3521)	(0.0491)	(0.3717)
AR (2): p-value	0.510	0.143	0.517	0.152	0.522	0.148	0.545	0.149
Hansen test: p-value	0.239	0.184	0.226	0.054	0.127	0.023	0.275	0.582
Instruments	18	18	18	18	18	18	18	18
Observations	4431	4431	4431	4431	4431	4431	4431	4431
Countries	130	130	130	130	130	130	130	130

*** p < 0.01, ** p < 0.05, * p < 0.1; robust standard errors are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

(Appendix A) show no evidence that overall financial development or its dimensions influence the top 10% income share or the Palma ratio, suggesting they do not affect income concentration at the extremes. One possible reason for the difference in estimation results between the top 10% share and the Gini coefficient is that percentile ratios do not utilize all available information, as they ignore incomes between the specified percentiles (Trapeznikova, 2019). Similarly, the Palma ratio considers only the extremes of the income distribution—the top 10% and bottom 40%—and ignores the middle half, making it a measure of income concentration rather than overall distribution, while being over-sensitive to changes at the extremes (Cobham & Sumner, 2013). It also does not capture inequality within the top 10% or bottom 40%.

Furthermore, our estimated overall financial development thresholds—around 0.34 in the fixed effects model and 0.60 in the system GMM model—vary across specifications, which may raise questions about regime stability and policy relevance. One potential reason for this discrepancy is that system GMM addresses endogeneity more effectively and captures the time persistence in the Gini coefficient, suggesting that the higher threshold of 0.60 is likely more reliable than the 0.34 obtained from the fixed effects model. To test threshold stability, we further perform sub-sample analyses by income level, as reported in

Table A4. Sub-sample analyses by income level. Effect of financial development, measured by FD, on income inequality: system GMM estimates.

	Low-Income Countries	Lower-Middle-Income Countries	Upper-Middle-Income Countries	High-Income Countries
	(1)	(2)	(3)	(4)
Gini (t − 1)	1.0995 ***	1.2640 ***	0.6517 ***	0.8472 ***
	(0.0377)	(0.2087)	(0.2374)	(0.0701)
Financial development (t − 1)	−0.0576	0.1466	−0.0171	−0.0621
	(0.0984)	(0.1562)	(0.0882)	(0.0530)
Financial development squared (t − 1)	0.4938	−0.3153	0.0482	0.0592
	(0.5085)	(0.3210)	(0.1149)	(0.0477)
Financial development (t − 1) × SBCD (t)	0.5407	−0.8629	−0.0040	0.0096
	(1.7127)	(0.6914)	(0.6697)	(0.1358)
Financial development squared (t − 1) × SBCD (t)	−2.8378	1.7696	0.0460	−0.0088
	(7.9222)	(1.4514)	(0.9626)	(0.1170)
GDP per capita, log	−0.0029	−0.0002	−0.0137	−0.0410 ***
	(0.0018)	(0.0076)	(0.0109)	(0.0151)
Openness	−0.0002	−0.0277	0.0156	0.0052
	(0.0076)	(0.0236)	(0.0205)	(0.0039)
Inflation	−0.0045	0.0015	0.0006	−0.0017
	(0.0045)	(0.0018)	(0.0005)	(0.0011)
Investment	0.0072	0.0171	0.0020	0.0310
	(0.0074)	(0.0320)	(0.0435)	(0.0386)
Government consumption	0.0169 **	0.0414	−0.1636	−0.0868
	(0.0079)	(0.0308)	(0.1087)	(0.0710)
FDI	−0.0166	0.0265	−0.1734	0.0133 **
	(0.0203)	(0.0367)	(0.1359)	(0.0059)
Human capital	0.0041	0.0089	−0.0008	0.0513 **
	(0.0030)	(0.0162)	(0.0187)	(0.0211)
SBCD	−0.0212	0.0903	−0.0001	−0.0013
	(0.0774)	(0.0722)	(0.0965)	(0.0344)
Constant	−0.0298	−0.1430	0.3077	0.3283 ***
	(0.0233)	(0.1008)	(0.1894)	(0.1214)
AR (2): p-value	0.585	0.416	0.473	0.424
Hansen test: p-value	0.162	0.608	0.520	0.565
Instruments	18	18	18	18
Observations	463	865	998	1477
Countries	22	31	34	43

*** p < 0.01, ** p < 0.05; robust standard errors are reported in parentheses; SBCD denotes the systemic banking crisis dummy.

in Appendix A. When estimating the model separately for low-, lower-middle-, upper-middle-, and high-income countries, neither the coefficient of FD nor its squared term is statistically significant. This indicates that the nonlinear effect of overall financial development on inequality is insignificant within these income groups, which may be attributable to the limited sample sizes and the restricted variation in financial development within each group, constraining the ability to detect nonlinear effects.

6. Conclusions

Using the IMF’s Financial Development Index and its sub-indices, this study examines the effects of financial development and its dimensions on income inequality. The results indicate that financial development initially reduces inequality but begins to exacerbate it once a threshold of around 0.34 (fixed effects) to 0.60 (system GMM) is surpassed. Since system GMM addresses endogeneity more effectively, the higher threshold of 0.60 is likely more reliable than the 0.34 obtained from fixed effects. Financial institution development exhibits a similar pattern, reducing inequality up to a threshold of approximately 0.56 (fixed effects) to 0.62 (system GMM), after which it increases inequality. The depth of financial institutions alleviates inequality at early stages but worsens it once the threshold—roughly 0.44 (fixed effects) to 0.60 (system GMM)—is exceeded. Likewise, access to financial institutions lowers inequality until about 0.64 (fixed effects) to 0.69 (system GMM), beyond which the effect turns positive. These effects are robust across different estimators and are not driven by systemic banking crises.

Our findings are consistent with the U-shaped relationship between financial depth and inequality proposed by Tan and Law (2012) and Cihak and Sahay (2020), as well as the U-shaped pattern between financial development and inequality identified by Brei et al. (2023). They also support Inoue’s (2024) argument regarding a U-shaped pattern between financial access and inequality. While the Literature Review discusses possible reasons why these effects turn positive, an additional explanation may relate to how credit is allocated: as credit expands, it may increasingly depend on collateral, firm structure, and industry sector—factors that tend to favor wealthier individuals and firms (Mbona, 2022).

Our study contributes to the literature by being the first to incorporate the IMF’s Financial Development Index and all eight of its sub-indices into both fixed-effects and system GMM frameworks to examine the nonlinear effects of financial development and its dimensions on income inequality. Unlike prior studies that primarily relied on indicators such as market capitalization to GDP or private credit to GDP, our approach provides a more comprehensive assessment by accounting for the depth, access, and efficiency of financial institutions and markets. Analyzing each sub-index separately allows for a more nuanced understanding of which specific dimensions of financial development mitigate or exacerbate inequality. The results show that the dimensions of financial institution development—particularly the depth and access to financial institutions—help reduce inequality when they remain below their respective thresholds but exacerbate inequality once those thresholds are surpassed. By contrast, other dimensions—the efficiency of financial institutions, financial market development, and its three sub-components—do not exhibit robust effects. These findings provide theoretical insights by demonstrating that the effect of financial development on inequality primarily driven by financial institution development—particularly its depth and access—rather than financial markets, with the effect, whether alleviating or aggravating inequality, depending on the level of these dimensions. It is also the first to combine the systemic banking crisis dummy, as per L. Laeven and Valencia (2020) and Nguyen et al. (2022), with the IMF’s Financial Development Index and its sub-indices, extending the analysis through 2019 and offering broader temporal coverage than previous studies.

Taken together, our findings have academic implications and contribute to the theoretical discourse. They provide new evidence that financial development, as well as its depth and access dimensions, exerts nonlinear effects on inequality, and these nonlinear effects are not driven by systemic banking crises. They also highlight the necessity of disaggregating financial development in order to better understand how it affects income inequality. Specifically, the results support our two hypotheses from the Literature Review section, showing that both financial depth and financial access exhibit U-shaped relationships with inequality. In the early phases of financial development, both affluent and disadvantaged groups benefit from financial institution development—particularly from greater access to and depth of financial institutions—thereby reducing inequality. However, as financial development advances, further financial deepening and expanded access disproportionately benefit those already well-off, ultimately exacerbating inequality.

Our study also has practical implications, providing guidance for policymakers on managing financial development to reduce inequality. Appendix B reports which countries, by income level, fall short of or surpass the estimated thresholds. Policymakers in low-income countries, as well as in middle- and high-income countries below the thresholds, should take action to promote financial development, especially by enhancing access to and depth of financial institutions, to help reduce inequality. In contrast, policymakers should shift toward curbing excessive financialization for middle- and high-income countries that exceed the estimated thresholds. In such cases, timely and well-targeted interventions to reduce inefficiencies in the financial sector are essential to mitigate inequality and promote social stability.

Our study has several limitations. First, the estimated thresholds for overall financial development differ across the fixed-effects and system GMM models, which may affect the robustness and policy relevance of the findings. We considered using the panel threshold model to identify regimes. However, this approach requires a balanced dataset, while our dataset is unbalanced with substantial missing values. Constructing a balanced dataset would necessitate dropping a large number of countries, thereby substantially reducing the sample size and limiting the ability to capture cross-country dynamics across different income levels. Moreover, there remains a gap in the empirical literature in applying Hansen’s (1999) panel threshold model to examine the effect of financial development on inequality using the IMF’s Financial Development Index and its sub-indices. Future research could address this gap by constructing a new dataset suitable for threshold analysis.

Second, although our study uses the most recent data available, extending the analysis through 2019 and thus providing broader temporal coverage than previous studies, it does not capture the period beyond 2019 due to data limitations. As a result, the potential effects of COVID-19 remain unknown. Once updated data become available, future research could re-examine our findings in light of the pandemic.

Third, while the IMF’s Financial Development Index and its sub-indices offer a more comprehensive measure of financial development than traditional indicators, they may still be subject to measurement limitations.