Green Banking Practices and Environmental Performance Mediated by Green Finance

2.1 Conceptual framework

This study examines the influence of green banking practices on the environmental performance of commercial banks (CBs) in Kathmandu by focusing on the mediating role of green finance. In this framework, green banking practices function as an independent variable, green finance as a mediating variable, and environmental performance as a dependent variable. The conceptual model illustrated in Figure 1 draws on prior research, particularly the work of Zhang et al. (2022), which provides a theoretical foundation for exploring these interrelationships.

Figure 1. Conceptual framework.

Conceptual framework illustrating the hypothesized relationships among Green Banking Practices (GBP), Green Finance (GF), and Banks’ Environmental Performance (BEP). The framework positions GBP as the independent variable, GF as the mediating variable, and BEP as the dependent variable, based on prior studies such as Zhang et al. (2022).

Source: Zhang et al. (2022).

2.2 Hypothesis formulation

Green Banking Practices (GBP) and Bank Environmental Performance (EP): GBP involves banks actively engaging in environmental sustainability efforts both within their operations and externally. Research indicates a strong link between GBP and improved EP in banks. For instance, Risal and Joshi (2018) found that green policies, environmental training, and energy-efficient technologies significantly enhance banks’ EP. Vidyakala (2020) also finds that adopting green banking practices positively influences environmental outcomes. These practices contribute to reducing environmental impacts by minimizing paper usage, fuel consumption, and carbon emissions, while fostering environmental benefits such as enhanced employee training, green infrastructure, and renewable energy utilization (Akter et al., 2017). Additional studies, including those by Shaumya and Arulrajah (2017) and Chen et al. (2022), further support the beneficial effects of GBP on EP.

Green Banking Practices (GBP) and Green Finance (GF): GF is a prime driver of sustainable development, emphasizing environmental stewardship and social responsibility (Liu et al., 2021). Green financing can effectively reduce both the internal and external carbon footprints of banks (Zheng et al., 2021). Chen et al. (2022) observe that various green banking activities, including those involving employees, daily operations, and institutional policies, positively influence green financing in Bangladeshi private commercial banks (PCBs). Rehman et al. (2021) further emphasized the significant influence of GBP on investments in green projects. Malsha et al. (2020) explored the connections between employee green behavior, GBP, and overall sustainability performance.

Green Finance and Bank Environmental Performance: Banks are pivotal in financing businesses and industries, which makes them responsible for environmental impact. Zhang et al. (2022) showed that GF positively influences the EP. Siddik and Zheng (2021) found that green financing helps reduce banks’ carbon footprint. Zhou et al. (2020) also reported that GF foster EP. Additionally, Bihari and Pradhan (2011) emphasized that GBP involves assessing the environmental consequences of projects before financing them, thereby promoting social and environmental accountability.

Green Banking Practices, Green Finance, and Banks’ Environmental Performance: The mediating role of GF in the link between GBP and EP is relatively under-explored. Zhang et al. (2022) document that GF serves as a mediator in this relationship within Bangladeshi commercial banks, amplifying the effect of GBP on EP. This suggests that GF acts as a catalyst, thereby enhancing the effects of GBP on environmental sustainability.

3.1 Research design

This study employed a quantitative causal research design to examine the effect of green banking practices (GBP) on banks’ environmental performance (BEP), with green finance (GF) as a mediating variable. Researchers have used a cross-sectional approach to collect data through structured questionnaires distributed both online and in the field. The design aimed to identify the direct and indirect relationships between GBP, GF, and BEP. Partial Least Squares Structural Equation Modelling (PLS-SEM) was used for data analysis, given its suitability for complex models and small sample sizes. This approach provides clear insight into how sustainable banking practices influence environmental outcomes.

3.2 Study area, population, and census

The research is centred on the Kathmandu Valley because of its concentration of corporate offices and headquarters for all 20 commercial banks in Nepal. As of mid-July 2024, these banks represented the full population of commercial banks operating in the country. Therefore, this study focuses on employees working within commercial banks in the Kathmandu Valley. This study collected data using self-administered questionnaires distributed during a field survey conducted between November 2024 and March 2025, ensuring timely and relevant responses from the target population.

Judgmental sampling was used in the first stage of this study. Employees working at the top ten profitable commercial banks in Nepal were taken as the sample frames. As per NRB in 2022, the banks that fall under the top ten profitable commercial banks of Nepal (in terms of net profit) include Global IME Bank Ltd. (GBIME), Rastriya Banijya Bank Ltd. (RBB), Nabil Bank Ltd. (NABIL), NIC Asia Bank Ltd. (NICA), Nepal Investment Mega Bank Ltd. (NIMBL), NMB Bank Ltd. (NMB), Nepal Bank Ltd. (NBL), Siddhartha Bank Ltd. (SBL), Prime Commercial Bank Ltd. (PCBL), and Kumari Bank Ltd. (KBL) respectively. Therefore, in the first stage, a judgmental sampling technique was used to select the top ten profitable commercial banks in Nepal.

In the second stage, a quota-sampling approach was applied. Quota sampling was employed to obtain data from commercial bankers in the Kathmandu Valley. A quota of 25 respondents from each of the selected top ten banks (based on net profit) was used, which adds up to 250 respondents as a sample size. Furthermore, to select 25 respondents from each of the selected top ten commercial banks, a convenience sampling approach was utilized. Although the study focused on a target sample of 250 respondents, with 25 respondents selected from each of the participating banks, an additional 29 samples were received, which were also used in data analysis as valuable information; 279 sample sizes were recorded.

Consent statement

Written informed consent was obtained from all participants after clearly explaining the study’s purpose, procedures, and right to withdraw at any time. No minor participants participated in this study. The Ethics Committee of Tribhuvan University, Faculty of Management, Nepal Commerce Campus, granted ethical approval for this study (Ref. No. 469/081–82) on 28 November 2024, before data collection began.

3.3 Variables and their measurement

The study used a five-point Likert scale to measure participants’ responses. On this scale, a rating of one represented strong disagreement, whereas a rating of five signified strong agreement with the statements associated with each variable. Tables 1 and 2 present the latent variables with their measurements, and the observed variables with their definitions, respectively.

3.4 Structural Equation Modelling (SEM)

The SEM is a sophisticated statistical method. This tool demonstrates the capacity to portray connections between diverse types of variables, encompassing both observable and latent variables, along with an increasing number of generated variables (Tarka, 2018). The SEM process involves a confirmatory method for analyzing data and is best utilized to confirm the validity of a particular model.

For an acceptable implementation, a higher level of advancement necessitates a more thorough understanding of facts and presumptions. Structural equation models extend traditional regression models by including many independent and dependent variables, which can be represented by collections of observable variables. The structural equation model encompasses two primary constituents: the dimension concept and SEM.

The measuring models specified in Muthén and Asparouhov (2012) are:

The SEM specified in the same is:

Where,

y = outcome variables

x = input variables

Λy = Observed response variables (latent variables)

Λx = Observed response variables (latent variables)

ε and δ = error

η = Latent variables (observed response variables)

ξ = Latent variables (observed response variables)

α = vector of intercepts

β = matrix of coefficients for endogenous variable regressions (ηi)

Γ = matrix of coefficients in the structural relationship of exogenous latent variables (ξ).

Within this context, the symbol “y” stands for the array of observable variables, while the symbol “x” signifies the array of input variables. Vectors δ and ε indicate the presence of measurement errors in the “x” (independent variable) and the “y” (dependent variable), respectively. Latent variables η and ξ were not directly observed in this study. Instead, the observed variables “y” and “x” were employed to assess the factor loadings Λy and Λx, respectively. These factor loadings illustrate the relationship between the observed and latent variables. Parameter α in the structural model represents a collection of intercepts. Matrix β comprises the coefficients governing the relationships between the endogenous variables (ηi), with zeros positioned on the diagonal. Matrix (I –β) remains nonsingular. The matrix Γ captures the coefficients governing the impact of the structural linkage of exogenous latent variables (ξ). Finally, ζ constitutes a random vector of the residuals.

However, the reduced form of the structural model in equations (1) through (3) can be written as follows if the errors are limited to the y-variables:

3.5 Data analysis

We analyzed the survey data collected via the KOBO Toolbox using MS Excel for data entry and tabulation, and SPSS and PLS-SEM version 4 for statistical analysis. PLS-SEM served as the primary tool for assessing the validity, reliability, hypothesis testing, mediation, and path analysis. The results are presented in the Tables and graphs. The analysis began with descriptive statistics to profile respondents, followed by inferential analysis to examine the impact of Green Banking Practices (GBP) on banks’ Environmental Performance (EP). The analysis was comprised of two main parts: descriptive and inferential.

4.1 Descriptive statistics

4.1.1 Socio-demographic characteristics of respondents

Sociodemographic characteristics of the surveyed sample are included in this section. Information was gathered from employees at 10 commercial banks in the Kathmandu Valley. These banks were selected based on their ranking as the top ten most profitable banks in terms of net income, with diversified portfolios (Karki et al., 2023), according to the 2022 annual report of the NRB. Furthermore, this section includes gender, age, bank, designation, education qualification, and working experience, which have been tabulated and presented in one kind of diagram so that a perfect picture of the socio-demographic characteristics can be interpreted.

Gender of respondents

Figure 2 shows that out of 279 respondents, there were 157 males and 122 females, which consisted of 56% and 44%, respectively. The researcher observed that the ratio of male respondents was higher than that of female respondents. The results of this study confirm that the majority of employees of the selected ten commercial banks in Kathmandu Valley are males, which shows that the banking sector seems to attract slightly more males than females.

Figure 2. Gender of respondents.

Distribution of the 279 respondents by gender, showing that 56% were male (157) and 44% were female (122). The data indicate a higher male participation in Nepalese commercial banks, reflecting trends noted in prior research that banking jobs attract slightly more male employees.

Source: Survey Data.

Similarly, a study conducted by Pande and Chalise (2022) stated that women occupy almost 40% of the workforce in the commercial (A-class) banking sector in Nepal. A similar study carried out by Khan et al. (2023) also found that 60% of the survey respondents who were commercial bankers were male and 40% were female.

Age group of respondent

Age is significant because of its linkage with accumulated individual and personal knowledge and behaviors. Figure 3 displays the age distribution of the survey respondents. The age groups were divided into four subcategories: 18 to under 28 years, 28 to under 38 years, 38 to under 48 years, and 48 years and above.

Figure 3. Age of respondents.

Age distribution of respondents across four groups: 18–28, 28–38, 38–48, and 48 years and above. The majority (57%) belonged to the 28–38 years category, highlighting the predominance of young and mid-career professionals in Kathmandu’s banking sector.

Source: Survey Data.

As shown in Figure 3, out of the 279 total respondents, the majority (57%) fell within the 28 to under-38 age group. The second largest group is those aged 18 to 28 years old, consisting of 25% of respondents. The 38 to 48-year-old group makes up 13.30% of respondents, and the 48-year-old and above group was the smallest at 4.70%. Hence, the data show that the age bracket of 28–38 years comprises the bulk of the responders.

Similarly, Tandukar et al. (2021) observed that most bankers were older than 21. Specifically, 22% of respondents were within the 31-40 age range, 4.9% were between 41 and 50 years, and only 2 percent were over the age of 51 years. Thus, the dominance of the 28 to 38 age group suggests that a significant portion of the banking industry’s workforce is filled with young and young adult workers in commercial banks.

Designation of respondents

The designation refers to the official job title given to employees. An individual’s designation at work can affect their ability to engage in green activities. In this study, the designation was divided into four parts: assistant, manager, officer, and other.

Figure 4 illustrates the distribution of the respondents by designation. The majority were assistants, comprising 36.92% of the sample, followed closely by officers at 36.20%. Managers represented 20.07% of the respondents, whereas other positions accounted for only 6.81%. This distribution indicates that most of the respondents were either assistants or officers.

Figure 4. Designation of respondents.

Distribution of respondents by job designation, showing that assistants (36.92%) and officers (36.20%) make up the majority, followed by managers (20.07%) and other positions (6.81%). The results suggest that survey participants largely represent the operational and mid-level strata of commercial banks.

Source: Survey Data.

Education qualification

The educational qualifications of employees considerably affect the implementation of GBP within organizations.

Figure 5 depicts the distribution of the respondents according to their level of education. The categories included Intermediate, Bachelor’s, master’s degree, and above. Out of 279 respondents, 67% held a Master’s degree, 28% held a Bachelor’s degree, and 5% had completed only the intermediate level. This indicates that the majority of the respondents possessed advanced degrees. Supporting this, Zhang et al. (2022) found that 5% had a Ph.D., 68.18% of respondents held a Master’s degree, and 27% had a Bachelor’s degree. This suggests that employees with master’s degrees and higher are predominantly engaged in commercial banks in Kathmandu Valley.

Figure 5. Educational qualification of respondents.

Respondents’ educational attainment, revealing that 67% held a Master’s degree, 28% a Bachelor’s degree, and 5% an Intermediate degree. The findings indicate that commercial banks employ a workforce with a high level of formal education, with postgraduate qualifications being dominant.

Source: Survey Data.

Working experience

An individual’s work experience demonstrates industry-specific knowledge, skills, and abilities. It provides an overview of their abilities and knowledge gained from their profession. Figure 6 shows the distribution of respondents by work experience, which is divided into four subgroups: below two years, two to four years, four to six years, and six years and above.

Figure 6. Working experience.

Distribution of respondents by years of work experience. Most respondents (40.14%) had six years or more of experience, followed by 27.24% with four to six years, and about 16% each with below two years and two to four years of experience. This reflects a mix of seasoned professionals and relatively newer entrants in the banking workforce.

Source: Survey Data.

Figure 6 illustrates that 40.14% have experience of six years and above, followed by 27.24% of employees who have experience of four to six years. An almost equal percentage is held by employees with work experience of two to four years and below two years, with 16.49% and 16.13%, respectively. Rehman et al. (2021) conducted similar research and found the following distribution of respondents according to their years of experience in the banking sector: In terms of years of experience, 5% of respondents had fewer than three, 35% had four to six, 40% had seven to nine years, 17.5% had 10 to 14 years, and 2% had 15 years or more of experience. In conclusion, it can be noticed that the majority of employees in the selected commercial banks have six or more experience of 6 years and/or above.

Respondent’s bank

Ten commercial banks were selected for this study. The commercial banks were sampled based on their ranking in terms of profitability (net profit) according to the annual report of 2022, as per the NRB. The banks selected according to that criterion are GBIME, RBB, NABIL, NICA, NIMBL, NMB, NBL, SBL, PCBL, and KBL. Although the study involved a target sample of 250 respondents, 25 respondents were selected from each of 10 participating banks, but an additional 29 samples were received. Therefore, 279 sample sizes were recorded.

General understanding of green banking practices

This section deals with the general understanding of GBP among employees of selected commercial banks in the Kathmandu Valley. The level of understanding of green banking practices can vary significantly among the employees. Figures 7 and 8 provide insights into how well employees in these banks understand and interpret green banking initiatives and their importance.

Figure 7. General understanding of the effect of GBP on lLong-term EP.

Respondents’ understanding of how Green Banking Practices influence long-term environmental performance. Nearly half (45.52%) perceived GBP as contributing to sustainable growth, while others viewed it as promoting behavioral change, mitigating climate risks, or even reducing workforce needs due to automation.

Source: Survey Data.

Figure 8. Understanding the importance of adopting ‘green banking practices’.

Respondents’ perspectives on why financial institutions should adopt GBP. The largest share (45.16%) emphasized mitigating climate risks and leveraging green investment opportunities, while others cited fulfilling social responsibility (25.09%), meeting regulatory standards (15.77%), and attracting environmentally conscious clients (13.98%).

Source: Survey Data.

Impact of green banking practices on the long-term environmental performance

In this section, the general understanding of the respondents regarding GBP and their influence on long-term EP is measured.

From Figure 7, it is obvious that 45.52% of the respondents have the understanding that GBP leads to sustainable growth. 27.96% of the respondents believed that GBP brings positive behavioral changes among people to protect the environment. Similarly, 14.7% believed that GBP was about mitigating climate-related risks. Lastly, 11.83% have the understanding that GBP will contribute to workforce reduction due to the automation of work.

Thus, it can be concluded that the majority of employees of commercial banks have recognized green banking’s role in fostering sustainable growth and positive behavioral shifts toward environmental protection. Additionally, respondents recognizing green banking’s potential to mitigate climate-related risks showcase an emerging awareness of the financial sector’s contribution to addressing environmental challenges.

Importance of financial institutions to adopt green banking practices

This section assesses respondents’ overall perceptions of the importance of financial firms adopting GBP.

The results are shown in Figure 8. It was found that almost half of the total respondents (45.16%) recognized the importance of financial institutions in adopting green banking practices to mitigate climate-related risks and capitalize on green investment opportunities. Followed by over a quarter of participants (25.09%) who agreed that the reason for accepting green banking initiatives was to fulfill the environmental and social responsibility of financial institutions. Furthermore, 15.77% and 13.98% think GBP is important to meet regulatory requirements and environmental standards, and to attract environmentally conscious customers, respectively.

The survey results demonstrated a moderate understanding among respondents regarding the importance of GBP for a variety of reasons. This understanding is highly driven by the desire to mitigate climate-related risks, capitalize on green investment opportunities, and fulfill the environmental and social responsibilities of banks.

This signifies a significant shift in the modern banking ethos. This indicates that these institutions are increasingly being viewed not just as economic entities but as stakeholders accountable for their environmental impact and societal contributions. The respondents’ identification of green banking as a strategic avenue for capitalizing on green investment opportunities underscores the alignment between environmental sustainability and financial profitability.

4.1.2 Research variables

Green Banking Practices (GBP)

Green banking practices (GBP), commonly referred to as green banking, involve environmentally conscious initiatives undertaken by banks to reduce their carbon footprints in daily operations. These practices also aim to minimize external carbon emissions (Sudhalakshmi & Chinnadorai, 2014). Table 3 illustrates the frequency and percentage of the total responses to statements related to GBP.

From Table 3 mentioned above, it can be seen that most of the people, i.e., 38.4% agree that their bank offers staff training on energy-saving equipment, environmental protection, and so forth. Following this, 41% of the respondents agreed that their bank organized seminars and meetings to promote eco-friendly practices. Likewise, 51.6% of respondents agreed that in the workplace, given the availability of alternatives, they prioritize environmentally friendly tasks whenever feasible, providing more advantages to their bank.

Similarly, almost half of the respondents (50.2%) show agreement that their bank attempted to reduce paper usage as much as possible. Additionally, 50.2% of respondents show agreement that their bank adopted energy-saving equipment such as eco-friendly ATMs, LED lighting, and eco-friendly printers. Furthermore, 36.9% of the respondents agreed that their bank used electronic waste management practices like recycling unused computers, monitors, old phones, etc. Likewise, 45.5% of respondents show agreement that their bank made investments in internet and mobile banking services. In addition to that, 45.2% of respondents agreed that their bank introduced innovative green products and services like green savings deposit accounts, green credit cards, and green OTP systems. Additionally, 47.3% of the respondents agreed that their banks had implemented green policies and pollution control plans. Finally, 41.2% agreed that their bank had green partnerships and agreements among suppliers and investors.

Most banks focus their investments on Internet and mobile banking services, with limited diversification in the equity market (Karki, 2018). Their priorities include adopting environmentally friendly alternatives and introducing innovative green products as part of their Green Banking Practices (GBP) to enhance environmental performance. Overall, the data show that the majority of employees of commercial banks have noticed the presence of GBP and perceive the banks as being moderately committed to adopting environmentally friendly initiatives. However, it is important to note that some responses were moderate or low in training and waste management, indicating a lack of priority in the GBP.

Green financing (GF)

Green finance (GF) focuses on renewable and alternative energy, waste management, green industry development, energy efficiency, recycling programs, and other green projects. The goal of GF is to promote organizational sustainability by incorporating environmental concerns into lending and investment decision-making, post-monitoring processes, and risk management, ultimately delivering environmental benefits (Siddik & Zheng, 2021).

From Table 4, it is evident that slightly more than half (51.61%) of the respondents agreed that their bank invested in renewable energy sectors like solar energy, hydropower, electric vehicles, and biomass energy. Likewise, 53.1% of the respondents agreed that banks offer financial products, such as loans, to companies that are seeking to make their operations more energy efficient.

Additionally, 43% of the respondents agreed that their bank invested (direct lending or own investment) in waste management projects. Moreover, 49.5% of respondents show agreement with the statement that their bank evaluates environmental risk for various projects and financing activities. On a similar note, 42.7% of the respondents agreed that their bank evaluated the environmental records of the clients while lending. Finally, 44.1% of the respondents agreed that their bank provided loans to eco-friendly projects with low interest rates and special grants (Bhandari et al., 2021).

Overall, the survey findings reveal that the respondents mostly expressed a moderate level of response when it came to GF in Commercial Banks. It can be noticed that banks are comparatively more actively adopting GF by investing in renewable energy, supporting energy-efficient projects, and assessing the client’s environmental records before lending. However, as per the respondents, it can be noticed that the banks have not been adopting GF to their full potential.

Banks’ Environmental Performance (BEP)

In this study, environmental performance denotes the effect of banking activities on the natural environment, focusing on reducing negative effects such as carbon emissions and waste, while enhancing positive outcomes through green banking practices for long-term conservation and business productivity (Shaumya & Arulrajah, 2017).

Table 5 reveals that the majority of respondents (61.3%) agree that their bank’s green banking practices enhance compliance with environmental standards. Additionally, 57.7% believed these practices help reduce energy consumption, while 48.8% agreed the bank promotes eco-friendly technologies. Half of the respondents (50.2%) recognized efforts to lower carbon footprints, and 53.7% agreed to reduce paper and material usage. Overall, 50.9% feel the bank evaluates the ecological impact of its activities. These findings suggest that respondents view green banking practices positively, particularly regarding environmental compliance and sustainability.

4.1.3 Descriptive analysis of research variables

Table 6 summarizes the respondents’ views on various Green Banking Practices (GBP), rated from 5 (strongly agree) to 1 (strongly disagree). The strongest agreement was with the assertion that banks focus on GBP through investments in online and mobile banking, with a mean value of 3.95, thus supporting Karki et al. (2021). Respondents also agreed that their banks offer innovative green products such as green deposit accounts and credit cards (mean 3.89). However, the lowest agreement (mean 3.59) was for the adoption of energy-saving equipment such as eco-friendly ATMs. Overall, mean values above 3 indicate general agreement with GBP initiatives, with a low standard deviation, reflecting consistent responses.

Table 7 shows respondents’ views on Green Finance (GF) practices. The majority agreed that their banks engage in GF through investments in renewable energy sectors like solar, hydropower, and electric vehicles (mean 3.63). They also agreed that banks assess environmental risks for projects (mean 3.62). However, there was a lower agreement (mean 3.42) on banks’ involvement in waste management investments. The neutral response to low-interest loans for eco-friendly projects (mean 3.47) reflects mixed opinions. Overall, the mean values above 3 indicate general agreement with GF practices, with low standard deviations, signifying consistent responses.

Table 8 summarizes respondents’ views on banks’ environmental performance (BEP). The highest agreement was with the statement that green banking practices enhance bank compliance with environmental standards (mean 3.86). The lowest agreement was regarding banks’ efforts to reduce their carbon footprint (mean 3.63). Overall, with mean values above three, most respondents agreed that their banks contributed positively to environmental performance. The standard deviations, ranging from 0.722 to 0.816, indicate steady responses with minimal variation from the mean.

In Table 9, it is evident that the overall mean value of green banking practices is 3.74, which surpasses the neutral point slightly. This suggests that among all ten statements related to GBP, respondents tend to agree that their respective banks have incorporated GBP into their operational framework. Furthermore, GF attains the lowest mean value of 3.54, illustrating that respondents opine that their respective banks do not significantly integrate GF through investment and lending practices. In contrast, BEP commands the highest mean value of 3.76, illustrating that the respondents show maximum agreement, with above-neutral mean value, that their respective banks contribute to environmental protection.

Inferential analysis

Inferential analysis seeks to interpret relationships between variables, evaluate differences, and make forecasts based on sample data using various statistical techniques (Kaur et al., 2018). It allows for generalizations about a population from sample data, supporting decision-making based on these insights (Simpson, 2015). This analysis includes the evaluation of the structural model, measurement model, and path analysis.

4.1.4 Measurement model results

The measurement model estimates the validity and reliability of constructs by examining internal consistency, reliability, convergent validity, content validity, and discriminant validity (Hair et al., 2014; Henseler et al., 2015). It is crucial to confirm a construct’s reliability and validity (Henseler et al., 2015). Cronbach’s alpha was used to assess internal consistency, while composite reliability should exceed 0.7, indicating sufficient reliability (Shrestha, 2021).

Table 10 shows the validity and reliability of the instruments employed in this study. Composite reliability (CR), average variance extracted (AVE), and outer factor loading were calculated. Factor loadings greater than 0.7 indicate adequate reliability (Hair et al., 2017). All AVEs exceeded 0.50, confirming strong item-construct relationships (Purwanto & Sudargini, 2021). The CR estimates were above 0.7, reflecting a high internal consistency (Shrestha, 2021). Additionally, the Cronbach’s alpha coefficients varied from 0.846 to 0.92, further demonstrating strong reliability and internal consistency across the variables.

Discriminant validity: Fornell and Larcker’s criterion

The degree to which a latent concept differs from other model constructs is determined by discriminant validity (Sujati et al., 2020). This confirms that the constructs represent separate entities, rather than overlapping concepts. This validity was examined using three criteria: Heterotrait-Monotrait Ratio (HTMT), Fornell-Larcker, and cross loadings. For any construct to meet the Fornell-Larcker criterion, which is frequently applied in PLS-SEM, the square root of its AVE must be greater than the correlations with other constructs (Almansoori et al., 2021). Table 11 shows the Fornell–Larcker criterion results.

Table 11 demonstrates that the square root of the AVE for each construct exceeded its correlations with other constructs, thereby satisfying the Fornell-Larcker criterion for discriminant validity. This confirms that the constructs measuring the impact of GBP on banks’ EP are unique and are not confounded by other variables.

Discriminant validity: Heterotrait-Monotrait Ratio (HTMT)

The HTMT ratio, another method for evaluating discriminant validity, assesses the distinctiveness of constructs. A common threshold is 0.85 (Roemer et al., 2021).

Table 12 reveals that all HTMT values are well below the threshold (0.85), confirming that the constructs related to GBP and banks’ EP are distinct and understood by the respondents.

Discriminant validity-Cross loading

Cross-loadings assess whether an item loads more heavily on its construct than others do. For valid discriminant results, an item should show higher loadings on its parental construct. Significant cross-loading (a difference of less than 0.10) suggests potential issues with discriminant validity (Ronkko & Cho, 2022), indicating that the item may not clearly distinguish between constructs.

Table 13 displays cross-loadings for three latent variables: GBP, GF, and Banks’ EP. As mentioned in Table 13, each item shows higher loadings on its parent construct than on others, indicating effective measurement of distinct constructs.

Structural model analysis

The structural model assesses the overall model fit and tests the research hypotheses (Alzoubi et al., 2020). It examines the complex relationships between observed and latent variables and evaluates measurement properties and structural interactions (Hair et al., 2021). In PLS-SEM, checking for multicollinearity using the Variance Inflation Factor (VIF) is essential because high multicollinearity can obscure the distinct contributions of independent variables.

VIF (Variance Inflation Factor) value

The VIF assesses multicollinearity among the study variables. When multicollinearity is high, it signifies a strong correlation between variables, which complicates the process of distinguishing between their individual effects on the latent constructs.

For this study, a VIF threshold of 3.3 is used to identify problematic multicollinearity. Values below 3.3 indicate no significant bias from multicollinearity. Table 14 shows that all the variables assessing the effect of GBP on banks’ EP fall below this threshold, suggesting no multicollinearity issues.

Path analysis

Path analysis, as shown in Figure 9, was used to evaluate the structural model by analyzing the direct and indirect relationships between variables (Fan et al., 2016). This technique visualizes hypothesized cause-and-effect relationships with arrows connecting variables, helping identify both direct and indirect effects. Figure 9 shows the path analysis of this study.

Figure 9. Path analysis results.

Structural model results from Partial Least Squares Structural Equation Modeling (PLS-SEM), showing the direct and indirect effects among GBP, GF, and BEP. Path coefficients, R² values, and factor loadings demonstrate significant positive relationships, with Green Finance partially mediating the link between GBP and BEP.

Source: Survey Data.

To examine the structural relationships between the constructs, three constructs were used: Green Banking Practices (GBP) with 10 items, bank environmental performance (BEP) with 6 items, and Green Financing (GF) with 6 items, totalling 22 items from 279 responses. Figure 9 shows the path coefficients, R² values, and factor loadings.

The path coefficients were 0.278, 0.369, and 0.339, indicating moderate positive correlations among GBP, GF, and BEP, with GBP and GF showing the strongest relationship. The R² values show that GBP explains 26.1% and GF explains 13.6% of the variance in BEP, demonstrating satisfactory model performance. All factor loadings exceeded 0.7, meeting the required threshold.

A path is statistically significant if its confidence interval does not include zero (Hair et al., 2021). From Table 15, the first hypothesis (H₁) is accepted, with a beta coefficient of 0.278 and a confidence interval ranging from 0.151 to 0.383, indicating that GBP significantly impacts banks’ EP. The second hypothesis (H₂) is also supported by a beta coefficient of 0.369 and a confidence interval from 0.247 to 0.473, confirming that GBP significantly affects Green Financing (GF). Additionally, the third hypothesis (H₃) is supported by a beta coefficient of 0.339 and a confidence interval between 0.219 and 0.44, demonstrating that GF significantly affects BEP. Overall, all hypotheses are supported, as their confidence intervals do not include zero, indicating that all relationships are positive and statistically significant.

Mediation analysis

Mackinnon et al. (2007) stated that mediation analysis seeks to understand the impact of a third variable on the relationship between two other variables. It assesses whether a mediator variable partially or fully explains the relationship between observed variables (Nitzl et al., 2016).

To examine mediation analysis, both direct and indirect effects must be considered. Tables 15 and 16 show that both the direct impact of GBP on banks’ EP and the indirect effect (GBP → GF → BEP) are significant because their confidence intervals do not include zero. This finding indicates that Green Financing (GF) partially mediates the relationship between GBP and BEP.