Financial technology and banking performance in developing countries: evidence from an advanced quantile regression approach

Empirical estimations

The estimates obtained using the econometric methodologies discussed in this section are offered here, along with a comprehensive breakdown and argument of the resultant statistics. Descriptive statistics offer comprehensive insights into the variables studied, and the results are reported in Table 3. BP shows a notable disparity between its mean of 3.035 and its standard deviation (SD) of 1.117. Notably, Iraq has the minimum BP value at −8.471, indicating its banks are among the least stable in the dataset. The other countries with the lowest BP values are the Syrian Arab Republic (SAR), Georgia, and Japan, which reflect higher financial risk for them. The maximum BP value, 3.450, was attributed to Kazakhstan in 2010, along with other top emitting nations, including Kyrgyzstan, Tajikistan, and Azerbaijan.

The FT mean is −3.505, with an SD of 1.317 values. Notable countries like Tajikistan, Pakistan, Cambodia, Nepal, and Iraq reflect lower values. Tajikistan reported a minimum value of −3.879 in 2011, indicating varied levels of FT. On the other hand, the United Arab Emirates (UAE), Japan, Thailand, and Kuwait consistently exhibit higher FT values, often surpassing 2. However, UAE achieved a maximum of 2.223 in 2017, highlighting commendable achievements and indicating strong digital financial service adoption.

The mean EG during the sample period is −8.165, with an SD of 1.402. Across most Asian economies, mixed-trend EG values are prevalent, with Japan, Singapore, Cyprus, and Malaysia reporting high values. Syrian Arab Republic records a minimum EG value of −6.349 in 2022. In contrast, Japan consistently demonstrates high EG values, reaching a maximum of 2.117 in 2015. The mean of OE is 0.172, with an SD of 0.791. The mean of CAD is 32.681, with an SD of 0.698. The mean of ING is 17.000, with an SD of 9.533, and the mean of GDP is 8.745, with an SD of 1.294.

The study evaluates the asymmetry of the numbers, specifically focusing on the non-normal distribution of the variables as indicated by their skewness and kurtosis values. In this examination, it is crucial to prioritize verifying the normal distribution of the statistics. Therefore, a normality test is employed (Jarque and Bera 1987). Consequently, it is experimental that all variables exhibit anticipated nonlinear patterns and refute the null hypothesis.

This study further employs two conventional methods to detect the presence of panel data issues, namely the tests for heterogeneity of slope coefficients and cross-sectional dependence. According to Tables 4 and 5, these assessments led to the following conclusions. The arithmetic significance of both the delta and the corrected delta was determined at the 1% level after examining the diversity in slope coefficients. Based on the findings of Hashem Pesaran & Yamagata, (2008), which rejected the null hypothesis of equal slope coefficients, it was determined that slopes exhibited non-uniformity across the distribution.

This observation affirms the distinctiveness of these economies in terms of their economic status, national political systems, power dynamics, monetary aspects, and other relevant factors. On the other side of the picture, CD evaluation results indicate that nearly all components have significant impacts. Hence, it may be deduced that the Developing economies exhibit characteristics of cross-sectional reliance. The CD test yielded noteworthy findings, particularly indicating the potential for interdependence among economies; any change in one economy could affect how healthy measures are done in other countries.

Table 6 presents the panel unit root test results using four approaches: Levin, Lin & Chu (LLC), Im, Pesaran & Shin (IPS), ADF-Fisher Chi-square (ADFF), and PP-Fisher Chi-square (PPF). These tests assess the stationarity of all variables employed in the analysis. The results indicate that all variables are stationary at level form across most testing methods at the 1% or 5% significance level. Specifically, BP, FT, CAD, GDP, and ING reject the null hypothesis of a unit root under all four test criteria, confirming their stationarity. However, EG displays mixed results. While the ADFF and PPF tests suggest significance at the 1% level, the LLC and IPS tests do not consistently reject the unit root hypothesis at conventional significance levels.

These mixed results for EG suggest that it may not be stationary in level form across all test methods. However, when first differenced, the variable achieves stationarity at the 1% significance level across all test types, indicating that it is integrated of order one, I(1). This finding implies that all variables included in the model are suitable for further panel data analysis, such as co-integration testing or dynamic panel estimation, as the majority demonstrate stationarity or become stationary after first differencing.

According to Table 7, the cointegration study yielded the following results. Consider a test statistic that is statistically significant at a 1% level for a cointegrated variable. In that case, it may be inferred that the variable is cointegrated. In the Asian setting, there are enduring associations among many financial indicators, namely BP, FT, EG, CAD, GPD, and ING. The confirmation of long-run cointegrating correlations is achieved using the suitable panel regression approach to predict long-run coefficients.

Empirical analysis outcomes

The results from the MMQR are presented in Table 8, providing detailed estimates across location, scale, and quantile-specific effects. The findings show that FT has a positive and statistically significant impact on BP across all quantiles, though the magnitude of this effect decreases at higher quantiles. Specifically, FT’s effect ranges from 0.135 (z = 2.09, p < 0.05) at the 10th quantile to 0.055 (z = 1.08, p < 0.10) at the 90th quantile. This pattern indicates that FT contributes more substantially to BP improvements in lower-performing banks compared to higher-performing ones.

This phenomenon may arise because FT provides financial goods and services to individuals and small firms without access to such resources. While the credit risk profiles of these new participants remain a concern, the results indicate that FT’s net impact is beneficial, potentially driving inclusivity without the destabilizing effects previously theorized. However, challenges persist, as regulatory and supervisory bodies in developing countries might face difficulties consistently keeping up with the rapid evolution of financial technologies. Ensuring that newly introduced participants in the financial system adhere to prudential norms and consumer protection requirements continues to pose significant challenges, particularly at higher quantiles where the impact of FT diminishes.

The study’s findings are partially consistent with those of Antwi-Wiafe et al. (2023), who showed that FT positively impacts African banks’ short-term and long-term performance. However, the new results suggest that the relationship between FT and BP is context-dependent, with FT providing net positive effects under specific economic conditions. Similarly, Nguyen et al. (2022) noted an adverse association between financial inclusion and performance, which may still apply in specific contexts but is less pronounced in the current study’s findings. The positive correlation between FinTech and bank profitability aligns with the view that FinTech credit can diversify financial portfolios, enhancing BP in emerging markets. However, these findings challenge the conclusions drawn by Dasilas and Karanović (2023), Yudaruddin, (2023), and (Kharrat et al. 2023), which suggested a uniformly positive relationship between FT use and improvements in BP.

The interaction term, FTEG, also shows a positive and statistically significant relationship with BP across all quantiles, although its effect is more modest. The coefficient is highest at the median quantile (0.097, z = 4.07, p < 0.01) and remains significant up to the 90th quantile (0.059, z = 2.09, p < 0.05). This implies that the effectiveness of FinTech is enhanced in countries with stronger governance structures, as sound regulation and accountability help mitigate risks associated with rapid financial innovation.

EG independently exerts a strong and statistically significant positive effect on BP throughout the distribution. At the 10th quantile, the coefficient is 0.288 (z = 4.08, p < 0.01), declining to 0.110 (z = 1.90, p < 0.10) at the 90th quantile. This result suggests that economic governance plays a more influential role in improving the performance of weaker banking systems, where institutional frameworks are essential for reducing systemic risk and ensuring confidence in financial operations.

Our findings align with those of (Ozili 2018), who determined that many factors, such as political stability, government efficacy, investor protection, regulatory quality, unemployment rates, and corruption control, play a crucial role in influencing banking stability in the African context. Study findings suggested that governance structures (GS) and law and order (LAO) have favourable and statistically significant effects on foreign policy (Danlami et al., 2023). Regarding control variables, all three, ING, CAD, and GDP, exhibit statistically significant relationships with BP. ING shows a consistent positive effect across quantiles, with a coefficient of 0.037 (z = 1.30, p < 0.10) at Q10, increasing to 0.062 (z = 2.75, p < 0.01) at Q90. This may reflect the pricing power of banks in inflationary contexts, particularly in developing economies.

CAD also has a strong and stable positive impact on BP, ranging from 0.211 (z = 3.26, p < 0.01) to 0.218 (z = 4.25, p < 0.01) across all quantiles, indicating that well-capitalized banks are better positioned to absorb risks and maintain performance levels. In contrast, GDP shows a statistically significant negative relationship with BP throughout the distribution. The coefficient declines from −0.251 (z = −4.81, p < 0.01) at the 10th quantile to −0.332 (z = −7.93, p < 0.01) at the 90th. This counterintuitive result may be due to the delayed transmission of macroeconomic growth benefits to individual bank performance in developing countries or possibly reflects increased competition and margin pressures during periods of economic expansion.

The empirical evidence from Table 8, along with the corresponding z-statistics and significance levels, confirms that FT, EG, and their interaction play crucial roles in shaping BP in emerging economies. These findings not only reinforce the positive impacts highlighted in earlier studies such as those by Antwi-Wiafe et al. (2023) and Ozili (2018) but also contribute to the ongoing debate regarding the potential risks and limitations associated with financial innovation in developing contexts (Nguyen et al. 2022; Dasilas and Karanović 2023; Yudaruddin 2023; Kharrat et al. 2023). Additionally, the significant effects of control variables like ING, CAD, and GDP provide further insights into how macroeconomic factors interact with technological and institutional variables to influence BP.

Discussion of results

The findings from our empirical analysis offer several important insights into the relationship between FT, EG, and BP in developing countries. The results indicate that FT is positively and significantly associated with BP across all quantiles. This effect is more pronounced at lower quantiles, i.e., the FT coefficient is approximately 13.5% at the 10th quantile, suggesting that financial technology may contribute more substantially to improving performance in relatively weaker banks. At higher quantiles, the impact declines to around 5.5%, indicating a diminishing marginal benefit of FT among higher-performing institutions. This pattern suggests that FT could play a facilitative role in enhancing financial inclusion by extending services to underserved populations and small enterprises. While some prior studies Antwi-Wiafe et al. (2023) and Nguyen et al. (2022) have reported negative associations between FT and BP, often attributed to credit risks or loan quality concerns, our findings underscore that these outcomes may vary by context. Differences in institutional environments, regulatory standards, and technological maturity likely influence the direction and magnitude of FT’s effects. In this regard, our findings are partially consistent with more optimistic perspectives presented by Dasilas & Karanović, (2023), Yudaruddin (2023), and Kharrat et al. (2023).

In parallel, EG is found to have a positive and statistically significant relationship with BP across all quantiles. For instance, a one-percent improvement in EG corresponds to a 28.8% increase in BP at the 10th quantile, with a gradual decline in effect size at higher quantiles. This suggests that robust governance frameworks, encompassing regulatory quality, rule of law, and institutional effectiveness, are especially critical for supporting banking performance in less stable financial environments. These findings are aligned with earlier work by Ozili (2018) and Hassan et al. (2023), who emphasize the importance of governance mechanisms in mitigating the risks associated with financial inefficiencies and corruption. Furthermore, the interaction term between FT and EG (FTEG) shows a modest yet statistically significant positive effect on BP, particularly around the median quantile. A coefficient of 0.097 at the 50th quantile implies that the complementarity between digital innovation and institutional strength can enhance bank performance. This reinforces the argument that FT’s benefits are maximized when embedded within a supportive regulatory and governance context.

The control variables also reveal noteworthy associations. ING has a positive effect on BP across quantiles, potentially reflecting banks’ ability to maintain pricing power in inflationary environments. CAD also shows a consistent and positive relationship with BP, highlighting the role of capital buffers in enhancing banking resilience. Conversely, GDP is negatively associated with BP at all quantiles, with coefficients ranging from −0.251 to −0.332. This counterintuitive result may reflect macroeconomic pressures or market competition in rapidly growing economies that constrain profitability. The findings suggest that FT may support banking performance, particularly among lower-performing institutions, but its effectiveness depends heavily on the quality of governance. Strong EG not only improves BP directly but also enhances the positive influence of FT. These insights contribute to the broader literature on the interplay between financial technology and economic governance, while also highlighting the importance of macroeconomic stability and sound regulation in shaping financial outcomes.

The robustness of these empirical results is further confirmed through Bootstrap Quantile Regression (BSQR), as illustrated in Table 9. The consistency between the MMQR and BSQR estimates reinforces the reliability of our findings and underscores the critical roles of both FT and EG in shaping bank performance in emerging economies.

Covid-19 (Covid‑2020)

The global outbreak of COVID-19 has adversely affected global economic conditions. In this study, Table 10 is used to examine the COVID-19 exogenous shock and its effects on the causative relationships between the level of banks’ pre-2020 FinTech expenditure and their performance throughout the progression of the pandemic. The present study reproduces the previously established results reported by (Ali et al. 2023; Phan et al. 2021; Shabir et al. 2022). Thus, the overall dataset, comprising 429 observations from 33 developing Asian countries over the period 2010–2022, was divided into three distinct subsamples. These subsamples represent the pre-COVID period (2010–2019), the COVID shock period (2020–2021), and the post-COVID recovery period (2022 onwards). The differences in subsample sizes arise solely from variations in the number of observations available each year and the total number of years included in each period. In other words, as the dataset is broken down by year, the total observations for each subsample reflect the data reporting practices and availability for that specific time window. The data presented in this study demonstrate a degree of resilience against the bearings of COVID-19. Research has shown that COVID-19 has severely impacted financial services due to the pandemic (Hassan et al. 2025; Phan et al. 2021; Shabir et al. 2022). With the COVID-19 pandemic factored in, the FinTech impact showed a consistent direction, statistical significance, and economic significance.

In both the pre-COVID-19 and during-COVID-19 periods, FinTech has consistently contributed to the positive performance of banks due to its presence. Notably, FinTech’s impact on bank performance was most apparent during Covid-19. After COVID-19, the current landscape has enhanced the benefits of FinTech for banks. Cutting-edge technology is one of the main factors contributing to the ongoing collaboration flanked by banks and FinTech businesses. In the post-pandemic period, it is imperative to bolster digital capabilities, optimize operational processes, and provide customer-centric services (Toumi et al. 2023). Analyzing COVID-19 results with post-adjustment procedures, utilizing panel data set at the country level.

Robustness analysis

To perform a robustness check, A further investigation was conducted using BSQR to assess the relationship between our crucial study variables, including BP, FT, EG, FTEG, CAD, ING, and GDP. We find that estimates of coefficients from all three approaches are consistent. The coefficient estimates in all models exhibit consistent magnitudes, signs, and significance levels. As a result, the results are more robust. FT, FTEG, CAD, and GDP stability variables significantly negatively correlate with BP over the long term. Meanwhile, CAD and ING have a positive effect on BP over time.

Panel causality outcomes

This analysis uses the Dumitrescu & Hurlin (2012) test to look for relationships between bank performance, FinTech, economic governance, and other control factors. As mentioned earlier, the technique has a causal relationship between these factors and practical qualities (this test can resolve the CSD problem). The causality is determined using the heterogeneous test non-causality (alternative hypothesis) and homogeneous non-causality (null hypothesis). Table 11 presents the test results from (Dumitrescu and Hurlin 2012).

The Dumitrescu-Hurlin panel technique was used in the current investigation to determine the causal relationship between each likely determinant. The cause-and-effect relationships between \({{FT}}_{{it}},{{EG}}_{{it}},{{FT}}_{{it}}* {{EG}}_{{it},},{{CAD}}_{{it}},{{GDP}}_{{it}},{and}{{ING}}_{{it}}\) are shown in Table 11. Banking performance and FinTech have a bidirectional causal relationship in Developing economies. Dumitrescu-Hurlin panel test results further demonstrated the existence of two-way causation between BPϕFT, BPϕEG, BPϕFTEG, BPϕCAD, BPϕING, and BPϕGDP, and BPυGDP. The bidirectional causality between BP and key predictors indicates a feedback loop where FinTech and governance reforms not only influence performance but are also shaped by it. However, GDP shows a consistent inverse causality, possibly reflecting structural lags in economic benefit transmission to bank-level outcomes.

Theoretical implications

This research has significant theoretical and practical implications for the study of FinTech, economic governance, and the performance of banks in developing countries. The prevailing positive assumptions about the impact of financial technology on banking need to be reconsidered in light of new evidence showing a positive correlation between FinTech development and bank performance. Additionally, our research underscores the importance of Asian financial institutions managing their FinTech resources effectively. This study further emphasizes the need to explore the complex relationships between legal frameworks, financial performance, and technological advancements. Gaining a deeper understanding of these connections can significantly enhance current frameworks and theories in finance.

Managerial implications

This study has important implications for financial industry professionals and decision-makers. First, it underscores the need for a balanced and sustainable approach to developing FinTech. Emphasizing responsible digitization and sustainability in FinTech initiatives is crucial for creating a strong and resilient financial landscape. Government oversight should be strengthened to ensure that FinTech initiatives are well-governed. Collaboration between various sectors is essential for establishing regulations that balance innovation with financial performance. Enhancing cybersecurity measures and enforcing strict data privacy regulations are vital for building trust in FinTech services and ensuring the stability of financial systems.

Interdisciplinary and societal implications

While our findings confirm the positive role of FinTech in improving bank performance, they also carry significant societal implications. In developing economies, the expansion of digital financial services is closely linked to broader efforts in reducing income inequality, enhancing financial inclusion, and bridging the urban-rural digital divide. The role of economic governance is particularly relevant in ensuring that FinTech does not exacerbate existing disparities, but rather supports equitable access to financial tools across socioeconomic groups.

Additionally, this study indirectly touches on emerging ethical considerations, including the use of AI and data-driven decision-making in banking systems. As FinTech becomes more integrated with artificial intelligence and algorithmic credit scoring, concerns arise regarding bias, transparency, and accountability in financial services. These questions underscore the need for not only sound governance frameworks but also ethical oversight in digital finance, a topic of growing interest in both economics and the humanities.