“Should income be diversified? A dynamic panel data analysis of Nepalese depository financial institutions”

This study analyzes the possible impact of diversity in non-interest income on Nepalese Depository Financial Institutions (DFIs) performance. The study examines variables such as service fees, dividends on equity instruments, and the non-interest revenue ratio to total operational income as endogenous factors. The ROE serves as the key profitability indicator. Additionally, the study explores the impact of control variables on the performance of financial institutions, such as the cost-to-income ratio, the equity-to-total assets ratio, and the ratio of non-performing loans to total loans. Secondary data from fiscal year 2015/16 to 2021/22 are utilized for analysis, employing correlation and regression analyses to assess the relationships between variables. Based on the Hausman Specification test, this study uses a Dynamic Analysis of Panel Data approach, adopting a Random effects regression model. The findings indicate that dividends from equity instruments ( β = –0.565*) adversely affect profitability. At the same time, service fees and non-interest revenue as a proportion of overall operating revenue show no significant impact. Control factors like the cost-to-income ratios ( β = –0.432**) and the equity-to-total assets ( β = –94.101**) adversely affect profitability. The study suggests that income diversification may not be beneficial, urging Nepalese DFIs to prioritize interest income and consider alternative investment opportunities. Reducing the cost-to-income ratios and equity-to-total assets is recommended for enhancing profitability.


INTRODUCTION
The financial sector is a cornerstone of any country's economic progress, and within it, a stable and prosperous banking industry is crucial for the growth of business and the economy as a whole (Arif & Nauman Anees, 2012; Karki et al., 2021).Traditionally, commercial banks have relied on net interest income from deposit collection and loan issuance as their primary revenue sources (Craigwell & Maxwell, 2006).However, lessons learned from the 2008 financial crisis emphasized the need for diversifying income sources to mitigate future risks (Prajapati & Shah, 2019).This shift is crucial to minimize over-reliance on interest revenue and to enhance overall financial stability (Antao & Karnik, 2022).Non-interest income (NII), which is defined as income from sources other than interest payments (Antao & Karnik, 2022), has gained prominence as interest margins narrow and banks seek revenue diversification (Heffernan, 2005).This strategic transformation is evident in the banking sector's transition from traditional lending activities to more fee-based services (Ghimire et al., 2023;Kumari, 2018).The banking landscape has evolved by intro-ducing various services, from credit cards to solvency certificates, creating a paradigm shift toward diversification (Andrzejuk, 2017).For instance, commercial banks charge multiple fees, such as non-sufficient cash, overdraft, and wire transfer fees (Dahal, 2022).The focus of international banks has changed from conventional to non-conventional revenue sources, aiming to boost profitability with enhanced earnings per share and dividend payout to their shareholders amidst intense competition, which leads to improved stock market performance (Karki, 2018).Deregulation, globalization, and technological advancements have been key drivers behind this shift (Craigwell & Maxwell, 2006;Dahal et al., 2020).Non-interest income now constitutes a substantial portion of operating revenue for US commercial banks (DeYoung & Rice, 2004).
Nepal's financial sector has undergone rapid structural changes since Nepal Bank Ltd. was founded in 1937 (Baral, 2007).The evolving landscape has highlighted Nepalese DFIs' need to adopt diversified income strategies.These institutions must consider how non-interest revenues affect profitability amid mergers, acquisitions, and heightened competition.Recent economic challenges have magnified the importance of diversification, including the COVID-19 pandemic, which disrupted economic activities and led to a decline in Nepal's Gross Domestic Product growth rate by 9.03% (World Bank).The economic downturn has prompted DFIs to reevaluate their revenue strategies.Furthermore, the regulatory requirement for Banks and Financial Institutions (BFIs) to maintain high credit-to-deposit ratios as part of economic stimulus efforts, coupled with challenges in retail loan processing under interest rate fluctuations and a liquidity crunch, emphasizes the importance of diversification (Bhandari et al., 2021).
The heavy reliance on interest income significantly affects BFIs' profitability trends and intensifies competitive pressures on margins.This has led researchers like Singh (2021) to emphasize the possible advantages of revenue diversification for bank stability.The reduction in the spread rate by Nepal Rastra Bank and the growing percentage of non-interest revenues highlights the shift towards non-traditional revenue sources (Dahal, 2022).In this rapidly evolving Nepalese financial market characterized by mergers, economic challenges, and competition, understanding the dynamics of revenue diversification and its impact on DFI performance becomes imperative.Valverde and Fernandez (2007) extensively examined the correlation between bank margins and specialty across seven European countries.Their study, from 1994 to 2001, revealed that market share and profitability improved significantly when European banks diversified into new business areas.This assertion supports the study by Chiorazzo et al. (2008), which examined the non-interest revenue and risk-adjusted returns of Italian banks over ten years.Their findings demonstrated a correlation between non-interest income and financial performance, suggesting that increased non-interest income leads to enhanced profits per unit of risk.

LITERATURE REVIEW
The dynamics of US credit unions were studied by Goddard et al. (2008) and revealed that heightened dependence on non-interest income positively correlates with enhanced profitability.Moreover, their research highlighted that a diversified revenue portfolio reduces the volatility of returns.Hong (2011) focused on the commercial banking sector in China and found a strong correlation between ROE and the non-interest revenue-to-business-income ratio.This underscored the pivotal role of non-interest revenue diversification in improving commercial banks' operational efficiency.Pennathur et al. (2012) analyzed how different ownership structures influence the earnings and risk profiles of Indian financial institutions.Their comprehensive study from 2001 to 2009 demonstrated that larger banks benefit significantly from higher non-interest income, whereas smaller banks exhibit more limited gains.Gurbuz et al. (2013) extended their analysis to the Turkish banking sector over six years and found that diversified revenue led to a better risk-adjusted performance.The literature review provides insights into key factors linked with income diversity influencing the profitability of financial institutions.The return on equity (ROE) ratio is an important performance metric that denotes the percentage of net profit to shareholders' equity and is a fundamental metric of an institution's financial well-being (Shah et al., 2018).It is widely assumed that a company's effective utilization of investments for growth is reflected in its ROE, a metric susceptible to various independent and control variables.Service fees, which include costs associated with account transactions, appear as a potential factor with significant implications for the profitability of financial institutions.Extensive research indicates a strong, positive association between service fees and financial performance, meaning that greater service prices correspond to higher profitability.Furthermore, dividends derived from stock instruments across various sectors favor a firm's profitability.This emphasizes the significance of income earned from such investments as a contributing component to the overall profitability of a financial institution.
Another important determinant is the Noninterest revenue to Total Operating revenue ratio, which reveals the importance of non-interest revenue in a bank's business model.

Prior research by Hong (2011) and Kozak and Wierzbowska (2022) indicates a correlation between this ratio and Return on Equity (ROE).
The ratio of Non-performing loans (NPL) to Total Loans, on the other hand, reflects the level of credit asset security within banks and provides a more dynamic relationship.However, Bhattarai (2016)  complexities of this relationship, demanding additional investigation.Another notable variable is the Equity to Total Assets Ratio, which serves as a metric of financial leverage.According to Prajapati and Shah (2019), this ratio positively links to profitability in the Nepalese context.Nepali (2018), on the other hand, shows contradictory findings, highlighting the necessity for a further examination of this association.Finally, the Cost to Income Ratio, which reflects the effectiveness of cost-cutting initiatives within banks, is significant.According to Sun et al. (2017) and Uddin et al. (2021), this ratio is inversely related to bank profitability, highlighting the importance of cost management efficiency.However, the appearance of contradictory conclusions in the literature emphasizes the complex dynamics governing these relationships.In this context, this research examines how endogenous and exogenous variables influence the financial performance of Depository Financial Institutions (DFIs), guided by the conceptual framework illustrated in Figure 1.
Based on the literature review, the following hypotheses can be developed.
H 1 : Service fees significantly impact the performance of DFIs.
H 2 : Dividends on equity instruments significantly impact the performance of DFIs.
H 3 : Non-interest income to total operating income significantly impacts the profitability of DFIs.
H 4 : NPL to total loan ratios have a significantly positive relationship with the performance of DFIs.
H 5 : Equity to Total Assets ratio significantly impacts the profitability of DFIs.
H 6 : Cost-to-income ratio significantly impacts the profitability of DFIs.

METHODOLOGY
This study utilizes a quantitative research methodology, combining descriptive and causal-comparative approaches.It focuses on how non-interest revenue impacts the profitability of Nepal's DFIs, comprising commercial banks, finance companies, and development banks.The study's sample spans seven years, from 2015 to 2022, encompassing various financial institutions.The sample consists of 10 DFIs, resulting in 70 observations.Purposive sampling, a non-probability sampling method, was used to choose the sample companies.Two criteria guided the selection process: first, the availability of consistent data over seven years, and second, the accessibility of data relevant to the study variables.The selected sample institutions are illustrated in Table 1.Global IME Bank (GBIME) 3.
Pokhara Finance (PFL) 10.Guheswori Merchant Banking & Finance (GMFL) The analysis employed secondary data from the NRB's website and publications.Descriptive and inferential statistics were applied to the analysis.The data were summarized and understood utilizing descriptive statistics like standard deviations and means.To assess the impact of independent variables (Service Fee, Dividend on equity instruments, and proportion of non-interest incomes) on the dependent variable (ROE), multiple regression and dynamic panel data analysis were employed that helped in testing the hypothesis and deriving inferential insights.Given the longitudinal nature of the data, panel data analysis was used to uncover trends over time.This approach was deemed suitable due to the study's focus on ten DFIs over seven years.The analysis was conducted using STATA 14.2, employing Pooled OLS, Fixed, and Random Effects regression models as estimation techniques.
Pooled OLS Model: This model, employing ordinary least squares regression, served as a baseline comparison.Based on Karki (2018), the model is represented by the following equation: where ROE it is the dependent variable representing ROE for institution i in year t.Independent variables are represented by SF it (Service Fee), DIV it (Dividend on equity instruments), and NONII it (Ratio of non-interest revenue to total operating income).Control variables include NPL it (proportion of non-performing loans), EQUITY it (proportion of equity to total assets), and CIR it (Cost-to-Income Ratio).α 0 is the regression constant, and β 1 to β 6 are the regression coefficients.μ i represents the time-invariant error, and ε it is the idiosyncratic error.

One-Way Fixed Effect Regression Model (FEM):
This model was chosen to elucidate how individual DFIs' specific characteristics might impact their profitability.This model allowed for a correlation between the explanatory and explained variables.
It was designed to address potential intercept variations across institutions due to their different non-interest revenue sources (Schmidheiny, 2022).
The equation for the one-way FEM was as follows: The term  (Schmidheiny, 2022).The equation for the two-way FEM was: Here tt T δ represented the time dummy variable, which accounted for the influence of temporal trends on profitability.Including both unit-specific and time-specific effects provided a more indepth comprehension of the variables shaping the performance of DFIs' over time.

Random Effects Regression Model (REM):
The study utilized the REM further to analyze the link between non-interest incomes and a bank's performance while accounting for potential confounding factors (due to the inclusion of many dummy variables).This model considered individual-specific effects uncorrelated with the explanatory variables, thereby mitigating the impact of omitted variable bias (Schmidheiny, 2022).The equation for the random effects model was:

RESULTS
This section illustrates and interprets the outcomes of various data analysis methods, comprising descriptive tests, correlation, and panel data analysis.Additionally, a negative correlation exists between the ratios of non-interest revenue to non-performing loans (NPL).This unexpected result contradicts the conventional assumption that non-interest revenue is related to improved performance.It suggests that higher levels of NPL might offset the positive effect of non-interest revenue on profitability (Sun et al., 2017).Further, the ratio of equity to total assets showed substantial negative correlations with three research variables: ROE, service fee, and dividend on equity instruments.This outcome underscores the complex relationship between equity ratios and various elements of financial performance, which has been observed in prior research (Nepali, 2018;Ngoc Nguyen, 2019).
Likewise, the cost-to-income ratio demonstrated a significantly negative correlation with the equity-to-total assets ratio at a 1% significance level.This result favors previous studies highlighting the impact of operating expenses on an institution's equity position and overall financial performance (Doumpos et al., 2016;Uddin et al., 2021).
Another noteworthy observation is the negatively significant correlation between the ratios of costto-income and non-interest revenue to total operating revenue at the 95% confidence level.This finding underscores the potential trade-off between controlling costs and generating non-interest revenue, as noted in prior research (Sun et al., 2017;Uddin et al., 2021).
The panel data analysis performed in this study aimed to explore the interactions between the dependent and various independent variables.Table 4 summarizes the findings derived from the Pooled OLS, FEM, and REMs.
The Pooled OLS regression model revealed an R-squared value of 57.51 percent, indicating that the independent variables explain approximately 57.51% of its variability in the defined variable.Furthermore, the significant F-value suggests an excellent overall fit of the model.Similarly, the Fixed Effects regression model yielded a significant F-value and an R-squared value of 46.24 percent.In the Random Effects  regression model, the R-squared value was 55.52 percent, with a p-value of 0.000 (<0.05), affirming the model's appropriateness and good fit.
Diagnostic Test: Diagnostic tests were carried out to identify the best model and ensure the reliability and quality of the data used for analysis.Mainly, they were conducted to assess multicollinearity, model selection, and heteroscedasticity, as detailed below: Variance Inflation Factor (VIF) Test: This test, as illustrated in Table 5, examined the occurrence of multicollinearity among the endogenous and control variables.The findings indicate that all variables showed VIF values below 10.The mean VIF value of 1.66, well below the threshold of 2, suggests that no significant multicollinearity exists in the model.This implies that the study variables are not strongly associated, confirming the suitability of the dataset for analysis.Hausman Specification Test: The Hausman Specification Test, detailed in Table 6, was employed to determine the suitable regression model between the FEM and the REM.The result, with a chi-square value of 4.03 and a p-value of 0.6723, shows that the null hypothesis -Random Effects Model is suitable -is accepted.This result supports the choice of the REM for the analysis.Furthermore, the regression coefficient for the ratio of non-interest income to total operating income (NONII) was 3.681, with a p-value of 0.598.The statistically insignificant p-value suggests that the proportion of non-interest incomes possess a positive but insignificant influence on the performance of Nepalese DFIs.This finding supports the assumption that non-interest income and profitability are positively correlated (Shah et al., 2018).
Regarding the control variables, the non-performing loan to total loan ratio (NPL) displayed a positive but insignificant impact on ROE, with a regression coefficient of 0.711 and a p-value of 0.277 (>0.05).In contrast, the equity-to-total assets ratio (EQUITY) demonstrated a negatively significant effect on the profitability of Nepalese DFIs, with a p-value of 0.000 (<0.05).The regression coefficient of the cost-to-income ratio (CIR) was -0.432, and its p-value was 0.000 (<0.05), suggesting a negatively significant impact on ROE.

P-value Remarks
H1: Service fees significantly impact the performance of DFIs.0.719 Rejected H2: Dividends on equity instruments significantly impact the performance of DFIs.
0.045 Accepted H3: Non-interest income to total operating income significantly impacts the profitability of DFIs.
0.598 Rejected H4: NPL to total loan ratios have a significantly positive relationship with the performance of DFIs.
0.277 Rejected H5: Equity to Total Assets ratio significantly impacts the profitability of DFIs.
0.000 Accepted H6: Cost-to-income ratio significantly impacts the profitability of DFIs.

Accepted
It is evident from the coefficients that the equity to total assets ratio exhibits the highest negative significance on ROE, followed by the dividend on equity instruments.These outcomes illustrate the complex interplay of various factors in influencing the profitability of DFIs, highlighting the need for an in-depth understanding of these relationships.

DISCUSSION
The study's primary finding highlights that non-interest revenue generally has an insignificant effect on the profitability of Nepalese DFIs.This contrasts with the results of In contrast, the sole independent variable with a significant impact on ROE was dividend on equity instruments, which exhibited a negative and significant impact.This finding contradicts Shah et al. (2018), who identified a positively significant relationship between dividend income and the performance of joint venture banks in Nepal.The variance in findings could be due to the severe economic penalties of the COVID-19 epidemic.The pandemic-induced economic slowdown prompted DFIs to invest in lower-yield securities, thus diminishing the positive effect of dividend income.This study also confirmed that the non-performing loan to total loan ratio had a positive and insignificant influence on ROE, consistent with Bhattarai's (2017) study.
Additionally, the equity to total assets ratio has a significant impact on ROE, aligning with Prajapati and Shah (2019).Consequently, Nepalese DFIs should consider the equity to total assets ratios as a determinant of ROE.The significantly negative effect of the cost-to-income ratio on ROE reso-

CONCLUSION
This study demonstrates that while non-interest income components, such as dividends on equity instruments and control variables, including cost-to-income ratio and equity-to-total assets ratio, significantly impact the ROE of Nepalese depository financial institutions (DFIs), other variables, such as service fee, non-interest incomes to total operating incomes ratios, and non-performing loan to total loan ratios do not.These findings hold valuable implications for decision-makers within Nepalese financial institutions, offering insights into income diversification strategies.The study highlights the need for developing a sustainable profit transformation model within the Nepalese banking system to reduce over-reliance on interest income and enhance income diversification, which is particularly crucial during financial crises.
provides a counterargument, suggesting that non-performing loan ratios have a positive influence on return on equity in Nepalese banks, contrary to the findings of research like Chiorazzo et al. (2008) and Doumpos et al. (2016).The discrepancy emphasizes the Investment Management and Financial Innovations, Volume 20, Issue 3, 2023 http://dx.doi.org/10.21511/imfi.20(3).2023.28

∑
represented the individual-spe- cific effects, capturing the distinct attributes of each institution.This model allowed researchers to assess how institution-specific traits interacted with various income sources to affect profitability.Two -Way FEM: It extended the insights from the one-way FEM by incorporating individual and time-specific effects.This model aimed to capture how personal attributes and temporal trends influenced DFIs' profitability nates with prior research by Doumpos et al. (2016), Sun et al. (2017), and Uddin et al. (2021).This emphasizes the importance of minimizing operating expenses and enhancing operating income for improved financial performance.

Table 3 .
Correlation matrixNote: ** One percent level of significance.* Five percent level of significance.

Table 5 .
Variance inflation factor

Table 8 .
Heteroscedasticity tests Sun et al. (2017)008)ho discovered a significantly positive relationship between non-interest revenues and the profitability of joint venture banks in Nepal.However, this inconsistency might be attributed to the inclusion of finance companies and development banks in this study, whereas Shah et al. (2018) focused solely on joint venture commercial banks.The diverse range of service fees among these institutions might explain the misalignment.Furthermore, the ratio of non-interest incomes to total operating incomes possessed a positive and non-significant impact on ROE.This result aligns with previous studies by Andrezejuk (2017), Antao and Karnik (2022),Goddard et al. (2008),Sun et al. (2017), and Tolangga and Ulpah (2019).These results suggest that a more significant proportion of non-interest incomes in total operating incomes can potentially boost the overall performance of Nepalese DFIs.