The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria

International Journal of Social Science and Human Research ISSN (print): 2644-0679, ISSN (online): 2644-0695 Volume 08 Issue 10 October 2025 DOI: 10.47191/ijsshr/v8-i10-24, Impact factor8.007 Page No: 7746-7759 IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7746 The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria Mary Ovayioza Ezeji1, Nasiru Abdulsalam Kao’je2, Yakubu Shaba3, Mohammed Auwal Babangida4 1,2,3Department of Accounting, Faculty of Management Sciences, Usmanu Danfodiyo University, Sokoto. 4Department of Taxation, Faculty of Management Sciences, Federal University Dutse, Jigawa State. ABSTRACT: Online transactions policy has gained significant traction over the years, with the increasing acceptability of electronic transaction channels such as Global Pay System (GPS), Corporate Internet Banking (CIB), Mobile Banking System (MBS), PayDirect System (PDS), and Remita (RMT). Therefore, this study examines the influence of online transactions policy on customers' deposits and bank operating expenses in Nigeria. Using mixed method sources of data and ex post factor research design, the SEMPLS models reveal that CIB (p = 0.000), GPS (p = 0.000), MBS (p = 0.029), and PDS (p = 0.000) strongly influence customers' deposits and operating expenses of DMBs in Nigeria. However, the panel regression models indicate that online transaction systems have no significant influence on customers' deposits, and only GPS, with a p-value of 0.021 and MBS, with a p-value of 0.095, increases operating expenses of the sampled DMBs. The study therefore suggests that improved online banking facilities, user education, and policy alignment are required to achieve the full benefit of online transactions policies in Nigeria. The study further recommends that DMBs should avoid a one-size-fits-all digital strategy. Instead, they should tailor digital tools to specific performance objectives such as deposit mobilisation, cost efficiency, and profitability. KEYWORDS: Online transactions, Customer deposits, Operating Expenses, Global Pay System, Internet Banking, Pay Direct System. 1. INTRODUCTION The Central Bank of Nigeria (CBN) launched the cashless banking (Online transactions) policy in 2012. On January 1, 2012, a cashless system policy test run began in Lagos state. The second stage of this strategy was implemented in Abia, Anambra, Kano, Ogun, Rivers, and the FCT on July 1, 2013, with statewide implementation on July 31, 2014 (CBN, 2019). This system is not intended to eliminate cash usage in transactions, but rather to limit physical cash handling and the amount of currency in circulation (Gbanador, 2021). The online banking system aims to increase the usage of electronic payment channels such as automated teller machines (ATM), point of sale (POS), mobile banking system (MBS), NIBSS instant payment, NIBSS electronic fund transfer, and other alternative payment channels. The overall impact of the cashless banking system on the performance of DMBs is complex and multifaceted (Okafor, 2020). Nigeria's online transactions (cashless banking) policy significantly improves business growth and GDP. However, certain disadvantages exist, such as inadequate information systems facilities, the possibility of fraud, high cost of IT facilities, high cost of running a business, and the perception of fraudulent activities linked to the cashless banking system. Nigeria has seen several cybercrimes and frauds over the years due to the cashless banking system's emphasis on electronic money transactions and the country's high degree of unreliable networks (Ogbeide & Fapohunda, 2017). Studies have been conducted on the influence of online transactions policy and the performance of DMBs (Makinwa, 2021; Okechukwu & Yua, 2021; Akani & Obiosa, 2020; Usman, 2020; Akindayo, Shadera & Solomon, 2020; Gambo, 2020; Nwakoby, Chukwu & Ogbenetega, 2020; Ul-Hug & Hossain, 2020; James & Eloho, 2020; Agu & Agu, 2020; Okafor, 2020; Ignoroje & Okoroyibo, 2020; Ogutu & Fatoki, 2019; Agu & Nwakwo, 2019). These studies have been characterised by inconclusive findings ranging from positive to negative and non-significant relationships. It has been observed that most of the studies have used variables such as ATM, Mobile Banking System (MBS), POS, CIB, and Electronic Fund Transfer (Akindayo, Shadera & Solomon, 2020; Gambo, 2020; Nwakoby, Chukwu & Ogbenetega, 2020; Ul-Hug & Hossain, 2020; Agu & Agu, 2020; Okafor, 2020) as proxies for online transactions policy. Therefore, this research extends the knowledge gap by addressing the influence of online transactions on customers' deposits and banks' operating expenses in Nigeria. The period covered by most of the studies reviewed (Akani & Obiosa, 2020; Usman, 2020; Agu & Agu, 2020; Okafor, 2020; Ignoroje & Okoroyibo, 2020; Ogutu & Fatoki, 2019) ends at the 2021 government calendar year, indicating that activities for the The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7747 years 2022, 2023, and 2024 remained uncovered. It should be noted that 2020, 2021, 2022, and part of 2023 were characterised by the global pandemic, which saw the restriction of banks' operations emanating from the social distancing and also cash crises in 2022 and 2023; hence the usefulness of the various cashless system platforms by the populace in Nigeria and around the world. Therefore, based on the foregoing, this research work studies the influence of online bank transaction systems on customers' deposits and banks' operating expenses in Nigeria, covering 13 years from 2012 to 2024. 2. CONCEPTUAL FRAMEWORK 2.1 Concept of Online Transactions Policy The Online Transactions Policy (OTP) refers to the conduct of financial transactions without the use of physical currency, leveraging instead credit and debit cards, telephonic and electronic fund transfers, the internet, and mobile banking. Cashless transactions occur using ATM, debit or credit cards, EFTs, or other media that do not require actual cash. Cashless banking is a system that aims to decrease the actual currency in circulation, without eliminating it, by promoting electronic transactions for payments, transfers, and other activities. (Kamboh & Leghari, 2016; Datta, 2021; Ernest & Fadiya, 2022). It signifies a transition period in the advancement of payment systems in most developing nations (Ikpefan, 2018). A cashless economy is the intermediate stage of a three-phase payment system economic model. This suggests that nations, particularly those in the development process, would shift from a mostly cash-based economic system to a cashless economy (Ighoroje & Okoroyibo, 2020). 2.2 Benefits of Online Transactions. The predicted patronage of online banking transactions system is expected to provide many advantages for various stakeholders. This provides users with more convenience, a broader array of service options, less vulnerability to cash-related crimes, and more economical access to banking services and credit beyond conventional branches. Businesses may benefit from improved accessibility, reduced revenue loss, and decreased expenses associated with cash handling (Atanda & Alimi, 2018); enhanced tax revenue, more access, and improved economic advancement for the government. A cashless banking system would enhance the standard of living by lowering the duration of transactions, boosting sales, simplifying cash collection, minimising transfer and processing costs, improving processing and transaction time, offering several payment options, and delivering instant notifications for all client account operations (Akhalumeh & Ohiakha, 2012; Andabai & Bina, 2019; Asenge, 2019). 2.3 The Challenges of the Online Transactions. The challenge of online transactions is that banks are vulnerable to various risks, such as transactional, strategic, reputational, and foreign currency threats. Lack of appropriate operational infrastructure, inadequate internet frameworks, and high expenses associated with acquiring and maintaining them sometimes prevent a significant portion of the population from using them, even in areas where they are available (Chison & Mike, 2018). 2.4 Customer Deposit Customer deposits refer to the money banks hold on their customers' behalf. These deposits are an essential element of a bank's obligations and serve as the foundation for the bank's lending operations. Various categories of deposits include savings, current, fixed deposits, among other deposits accounts. Deriving certain pivotal indicators on client deposits might provide valuable insights into a bank's performance and financial well-being. 2.5 Operating Expenses Operating expenditures in DMBs refer to the various costs of managing the bank's daily operations. These expenditures are essential for continuing banking services and maintaining a competitive position in the financial market. Typical operational expenditures in DMBs include compensation for bank workers, including salaries, wages, bonuses, and perks, which constitute a substantial proportion of operational costs. Expenses associated with renting or leasing office spaces, branches, and other facilities are part of the operational costs for DMBs. Operating expenses include the expenditures related to utilities, such as power, water, and heating or cooling systems, at bank branches and offices. Expenses related to the upkeep and enhancement of banking systems, software licensing, and technical infrastructure are crucial elements of operational costs. Operating expenditures are increased by the costs incurred in maintaining security measures, which include staff, surveillance systems, and security technology. DMBs may be subject to charges for professional services such as legal counsel, consultancy, and auditing, contributing to operational costs. The routine upkeep and necessary fixes of tangible assets, such as ATMs, buildings, and equipment, add to the operational expenses of banks. 2.6 Empirical Review and Hypotheses Development A comprehensive investigation has been carried out on the nexus between the cashless banking system and the growth of DMBs in Nigeria. However, there is a significant lack of research that combines several elements to measure performance. Examples, Makinwa (2021), Okechukwu and Yua (2021), Akani and Obiosa (2020), Usman (2020), Akindayo, Shadera, and Solomon (2020), The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7748 Gambo (2020), Nwakoby, Chukwu, and Ogbenetega (2020), Ul-Hug and Hossain (2020), James and Eloho (2020), analyse the impact of a cashless policy on DMBs performance in Nigeria. Adu and Williams (2022) investigates the effects of cashless policies on the profitability of Nigeria DMBs. The objectives encompass examining the influence of ATM, NEFT, POS, and e-banking transactions. The research indicated that ATM, NIP, mobile banking, and cheque transactions significantly influenced the performance of the sampled DMBs. The report recommended banks to fund training for technical personnel overseas to acquire knowledge and stay current with new and innovative technology utilised in the banking industry. Adamu and Adaora (2023) examine the effect of inclusion initiatives on the performance of Nigeria's banking industry. Using an ex-post facto technique, the OLS model reveals that POS and ATM substantially impact system's efficiency and effectiveness. The research recommends, among other things, that the CBN should reduce electronic payment transaction rates to promote broader inclusion. Charity, Victoria, Chima, and Udeoba (2024) explore the impact of ATMs and mobile banking on the performance of Nigerian DMBs. The analysis used data from the CBN Bulletin from 2009 to 2021. The findings indicates that ATMs have a significant impact on the performance of Nigerian DMBs. The study concluded that e-banking has enhanced banking access for users while also allowing banks to extend their operations to reach more people. Ajuonu and Uzodike (2024) investigate the impact of the Cashless Policy on the Accounting Performance of Some Selected DMBs in Nigeria. The ex-post facto research design was used, and the data were analysed using multiple regression analysis. The study discovered that ATM transactions substantially affected the ROE of DMBs in Nigeria. However, POS transactions significantly affect the NPM of selected DMBs. The study concluded that the volume of ATM and E-banking transactions substantially impacts Nigerian DMBs' revenues, as does the volume of POS transactions on client deposits. Prior research, including Agu and Agu (2020), Okafor (2020), Ignoroje and Okoroyibo (2020), Ogutu and Fatoki (2019), Agu and Nwakwo (2019), Itah and Ene (2014), and Abaenewe, Ogbulu, and Ndugbu (2013), has frequently employed metrics such as ROE, ROA, EPS, and NPAT. Nevertheless, there exist inconsistencies in their findings. Owing to these controversies in their findings, the study hypothesised that: H01: Online transactions policy does not significantly impact customers' deposits of listed Deposit Money Banks (DMBs) in Nigeria. H02: Online transactions policy does not significantly impact the operating expenses of the listed Deposit Money Banks (DMBs) in Nigeria. 3. RESEARCH METHOD 3.1 Research Design The study used a mixed-methods research strategy, using an ex post facto approach. The ex post facto research design, or a retrospective or non-experimental research design, is an observational study where the researcher examines the relationship between variables without actively manipulating them. The study population comprises nineteen (19) listed commercial banks classified into national and international licensed banks on the Nigerian Exchange Group (NGX) as of 2021. Census sampling techniques were employed since the entire population was studied. 3.2 Sources of Data Data was sourced via structured/closed-ended questionnaire and annual accounts and reports of the sampled firms. A five-point Likert-type scale was used to seek a response rate and reduce respondents’ frustration level, as noted by Pondent (2017). In addition, a structured question is mostly recommended due to the desire to avoid a lacklustre response, which could result in common method bias (Dillman, Smyth & Christian, 2014). The data was analysed with the aid of SEM-PLS and panel regression techniques. The study covers 12 years, spanning from 2012 to 2023. The justification for selecting this time frame is based on the fact that the cashless system was introduced on January 1 2012, and since then, it has occupied a major policy applicable in the Nigerian banking sector. 3.3 Methods of Data Analysis The study utilises both inferential and descriptive statistics for data analysis. The regression analysis was employed to analyse the quantitative data and further show the significance level of the variables, from which a stance was made on the failure to reject or reject the earlier formulated hypotheses. In addition, the study used a structural equation model, specifically Smart-PLS, to analyse the qualitative data and ascertain the research instrument's validity and reliability. These were undertaken after preliminary analyses, including missing values, nonresponse bias, common method bias, detection of outliers, and normality tests using SPSS version 27. 3.4 Instruments of Data Analysis The instruments used in this study were adopted from past studies, with minor changes to suit the study. Firm performance was measured using operating expenses and customer deposits adopted from Spillan and Parnell (2006). The online transactions were measured using the adopted 7-item Adamson University Survey Questionnaire on Cashless Society (2018). A sample of a question The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7749 is “Online transactions are much more convenient". The reported Cronbach's alpha is 0.798, showing the instrument is reliable and adequate for the study. 3.5 Model Specification The panel regression model to test the hypotheses was adopted from the study of Chikwemme and Nwadialor (2019). The models were specified as follows: FP = OT…………………………………………………..…………………………………….………………… (i) Where the equation provided reads thus: FP = β0+β1CIBit+β2MBSit+β3GPSit+β4PDSit +β5RMTit+Ɛit ………………………….…….(ii) OE = β0 + β1CIBit + β2MBSit + β3GPSit + β4PDSit + β5RMTit + Ɛit ….........................(iii) CD = β0 + β1CIBit + β2MBSit + β3GPSit + β4PDSit + β5RMTit + Ɛit………….…….........(iv) Where: OT = Online Transactions GPS = Global Pay System CIB = Corporate Internet Banking MBS = Mobile Banking System PDS = Pay Direct System RMT= Remita FP = Financial Performance CD = Customer Deposit OE = Operating Expenses β1 - β5 = The parameters to be estimated i = Represents individual firms t = Stands for time period ε = The error term of the model 3.6 Variables Description and Measurement This sub-section describes the study's various variables, which are made up of both dependent and independent variables, and how they are measured (proxies) as presented in Table 1. Table 1: Variables’ Description and Measurement Variables Measurement Prior Studies Dependent Variables = Firms’ Performance Customers' Deposit (CD) Log of total operating expenses of sampled DMBs Hussein and Elyjoy (2018), Eminah (2022) Operating Expenses (OE) Log of customers' deposits Emenike (2020) Independent Variables: Cashless System Global Pay System (GPS) Total volume of global pay transactions for DMBs Very few studies have employed this variable to measure the cashless system. Corporate Internet Banking (CIB) Total volume of corporate internet banking transactions for DMBs Ihpefan et al. (2018), Mustapha (2018), Oyomo (2018), and Okafor (2020). Mobile Banking System (MBS) Total volume of mobile banking transactions for DMBs Ihpefan et al. (2018), Mustapha (2018), and Oyomo (2018). Pay Direct System (PDS) Total volume of direct pay transactions for DMBs Ene (2020) Remitta (RMT) Total volume of Remitta transactions for DMBs Ene (2020) Source: Author’s Compilation (2024). 4. Data Presentation and Analysis This sub-section mirrors data collected by administering questionnaires to respondents, and the computation is provided in Table 2. The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7750 Table 2: Questionnaire Distribution and Response Rate Source: Field Work (2024). The distribution and response rate of the questionnaires for the study comprised 422 questionnaires, 417 of which (98.8%) were returned. Among the returned questionnaires, 15(3.6%) were rejected or removed. Consequently, 402(95.2%) were retained and deemed usable for analysis. This high response rate and retention of usable questionnaires indicate a robust sample size, enhancing the reliability and validity of the study's findings. This is in tandem with Ryu (2020) and Ojeleye, Abu-Abdissamad, Umar, and Usman (2022) assert that a consensus on a minimum acceptable response rate is lacking. Malhotra and Grover suggested that a 50% response rate is sufficient for data processing and presentation. 4.1 Analysis of Missing Values Table 3 presents the analysis of the missing values. Table 3: Computation of Missing Values Latent Variables No. of Missing Values Global Payment System 6 Corporate Internet Banking 5 Mobile Banking System 7 Pay Direct System 5 Remita 3 Financial Performance 8 Total 34 out of 12,060 data points Percentage of missing values: 0.28% Source: Researcher’s Compilation (2024). Note: The missing value percentage is derived by dividing the number of randomly missing values for the entire data set by the total number of data points multiplied by 100. Table 3 shows that only 34(0.28%) data points from the initial data set were randomly missing. However, no "golden rule" exists for an acceptable percentage of missing values for valid statistical inference in a data set (Aliyu, 2020). Researchers have established that missing data of less than 5% for the sample and 10% for the variables are generally acceptable and agreeable to any imputation strategy (Hair et al., 2014). Since the missing data was less than 10%, the mean substitution imputation approach replaced the missing data. Furthermore, experts believe that mean substitution is the simplest technique for replacing missing data with a probability of less than 5% (Tabachnick & Fidell, 2012). 4.2 Normality Test Studies (Wetzels, Odekerken-Schröder, & Van Oppen, 2009; along with Reinartz, Haenlein, & Henseler, 2009) have established that SEM-PLS functions effectively with non-normally distributed data. Similarly, Hair et al. (2019) posited that highly skewed data elevates bootstrap standard error, perhaps resulting in an underestimation of the statistical significance of route coefficients. Therefore, this study employed multivariate normality to analyse the data distribution by utilising Kurtosis and Skewness to assess normalcy. Table 4 presents the ratings for Skewness and Kurtosis of the research variables. Table 4: Normality Test: Skewness and Kurtosis Statistics (n=412) N Stat. Min. Max. Mean Std. Dev Skewness Std. Error Kurtosis Std. Error CIB 402 1.235 5.000 3.743 0.484 1.031 0.122 0.332 0.244 GPS 402 1.342 5.000 3.943 0.348 0.238 0.122 -1.363 0.244 MBS 402 0.987 4.958 3.854 0.389 0.680 0.122 -0.520 0.244 PDS 402 1.123 4.896 3.847 0.386 0.398 0.122 -1.014 0.244 FP 402 1.286 5.000 2.765 0.904 0.702 0.122 -0.725 0.244 RMT 402 1.243 5.000 2.515 0.612 0.322 0.122 -0.459 0.244 Source: SPSS Version 27, (2024). Questionnaire Frequency Rate% Distributed 422 100 Unreturned 5 1.2 Returned 417 98.8 Rejected 15 3.6 Retained 402 95.2 The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7751 Based on the analysis in Table 4, it is determined that the data is normal because the absolute values of Skewness and Kurtosis for all items in this research are within the permitted ranges of < 2 and < 7, respectively. Skewness levels of less than 2 and Kurtosis values of fewer than 7 are advocated (Curran & Blackburn, 2001). Kline (2016) contends that absolute Skewness values exceeding 3 and Kurtosis values surpassing 10 may indicate a concern, however values over 20 may denote a more serious issue. 4.3 Multicollinearity Test Two (2) techniques are used in this work to detect multicollinearity. The correlation matrix of the exogenous latent constructs functioned as the initial approach. A correlation coefficient of 0.90 or above signifies multicollinearity among exogenous latent variables (Hair et al., 2019). Table 5 demonstrates that the correlations among all exogenous variables were beneath the threshold (i.e., r = 0.90). The correlation matrix demonstrates the lack of multicollinearity among these variables. Secondly, the researcher looked at all the exogenous latent constructs' tolerance levels and variance inflation factor (VIF) values. Hair et al. (2014) and Field (2017) advocated that the commonly recognised level of multicollinearity of a VIF value ≤ 10 indicates no multicollinearity, but a value more than 10 indicates severe multicollinearity. While a tolerance level of ≥ 0.1 indicates no multicollinearity, a number less than 0.1 indicates substantial multicollinearity. However, Kock (2015) argued that the presence of VIF more than 3.3 and less than 0.303 tolerance level is a sign of pathological collinearity and common method bias. Table 5: Correlation Matrix (n=402). Construct GPS CIB MBS PDS RMT GPS 1 CIB .031 1 MBS .681** 0.021 1 PDS .485** 0.021 .732** 1 RMT .812** 0.000 .553** .553** 1 Source: SPSS Version 27 (2024). **. Correlation is significant at the 0.01 level (2-tailed). Table 5 shows that multicollinearity is absent among the exogenous latent components since the VIF values are smaller than the 0.303 threshold Kock (2015) indicated. As a result, multicollinearity is not a concern in our investigation. Table 6: Tolerance Level and Variance Inflation Factor (VIF) Values Collinearity Statistics Endogenous Variable Latent Constructs Tolerance level VIF GPS 0.836 1.196 FP CIB 0.996 1.004 MBS 0.502 1.992 PDS 0.389 2.574 RMT 0.485 2.062 Source: SPSS Version 27 (2024). Table 6 shows the VIF values for five latent constructs, assessing collinearity. A higher tolerance level (>0.2) and VIF values below 5 indicate low multicollinearity, suggesting reliable data for regression analysis. GPS (Tolerance: 0.836, VIF: 1.196) and CIB (Tolerance: 0.996, VIF: 1.004) have very low VIF values, indicating minimal multicollinearity. 4.4 Assessment of SEM-PLS Path Modelling The statistical outcomes of the various analyses, such as the reliability of individual items assessing each latent concept, internal consistency, and discriminant validity, among other factors, were examined. Therefore, the PLS algorithm was used to assess individual item reliability and other measurement model criteria, with outer loadings indicating the reliability of reflective constructs. According to Hair et al. (2014), an outer loading of 0.70 is ideal for established scales, but items with loadings between 0.40 and 0.70 may be retained if leaving them improves the AVE and CR. Following Hulland’s (1999) guideline, loadings below 0.5 should be excluded. Consequently, five items were removed, improving the AVE and CR, leaving the remaining items suitable for further analysis as shown in Figure 1 and Table 6. The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7752 Figure 1: Measurement Model Source: Smart PLS version 3.2.8 (2024). Figure 1 and Table 6 collectively present the measurement model's assessment, showcasing the evaluation of both item reliability and the overall construct validity and reliability for the reflective constructs examined in the study. The indicator loadings for each construct met the minimum threshold for acceptability, indicating sufficient individual item reliability for most items. However, five items—FP3, FP5, FP8, GPS3, and RMT1—fell below the acceptable loading criteria and were removed to enhance the model's quality. Eliminating these items improved AVE and CR metrics, strengthening construct validity. The remaining items, all of which met or exceeded the required loading threshold, were deemed reliable and valid for continued analysis in the study. This refinement ensures that the measurement model aligns with established academic standards for reliability and validity, enabling more robust and accurate structural model evaluations in subsequent analyses. Table 7: Item Loadings, Internal Consistency, and AVE. Constructs Indicators Loadings Cronbach's Alpha Composite Reliability AVE Corporate Internet Banking CIB1 0.803 0.781 0.858 0.603 CIB2 0.692 CIB3 0.779 CIB4 0.826 Global Pay System GPS1 0.807 0.715 0.812 0.521 GPS2 0.642 GPS4 0.683 GPS5 0.745 Mobile Banking System MBS1 0.778 0.827 0.877 0.588 MBS2 0.748 MBS3 0.801 MBS4 0.839 MBS5 0.657 Pay Direct System PDS1 0.879 0.762 0.847 0.585 PDS2 0.830 PDS3 0.711 PDS4 0.609 The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7753 Financial Performance FP1 0.724 0.785 0.854 0.540 FP2 0.721 FP4 0.805 FP6 0.655 FP7 0.759 Remita RMT2 0.738 0.705 0.765 0.526 RMT3 0.838 RMT4 0.576 Source: Researcher’s Composition (2024). 4.5 Assessment of the Structural Model This section presented the SEM for data analysis on direct and mediated relationships through bootstrap analysis. A typical bootstrapping method was employed, utilising 5000 bootstrap samples from 402 examples to examine the importance of the route coefficients in the direct relationship (Hair et al., 2014; Hair et al., 2021). Figure 2: Structural Model Source: SEM-PLS Version 3.2.8 (2024). Table 8: Path Coefficient (Direct Effect) Hypotheses R.Ship Beta Std. Dev. T Statistics P Values Decision H01 GPS -> FP 0.374 0.038 9.842 0.000 Rejected H02 CIB -> FP 0.189 0.037 5.031 0.000 Rejected H03 MBS -> FP 0.096 0.044 2.188 0.029 Rejected H04 PDS -> FP 0.336 0.031 10.683 0.000 Rejected H05 RMT -> FP 0.011 0.040 0.289 0.773 Accepted Source: SPSS Version 27 (2024). ***p<0.01, **p<0.05. Table 8 presents path coefficients representing the direct effects on licensed DMBs' performance in Nigeria. The first hypothesis (H01) posits a relationship between Global Pay System (GPS) and financial performance (FP), with a beta coefficient of 0.374. The standard deviation (STDEV) associated with this coefficient is 0.038, yielding a T statistic of 9.842 and a p-value of 0.000 at a 1% (p<0.01) significance level. The null hypothesis is rejected, suggesting that GPS significantly influences the performance of License DMBs in Nigeria. The second hypothesis (H02), which examines the association between Corporate Internet Banking (CIB) and FP, exhibits a beta coefficient of 0.189. The associated STDEV is 0.037, resulting in a T statistic of 5.031 and a p-value of 0.000. Consequently, H02 is rejected, indicating a significant impact of CIB on the FP of licensed DMBs in Nigeria. The third hypothesis (H03) focuses on the Mobile Banking System (MBS) and its effect on FP, revealing a coefficient of 0.096, a standard error of 0.044, a t-statistic of 2.188, and a p-value of 0.029. These figures indicate significance at the p<0.05 level. The Influence of Online Transactions on Customers Deposit and Operating Expenses of License Deposit Money Banks (DMBs) In Nigeria IJSSHR, Volume 08 Issue 10 October 2025 www.ijsshr.in Page 7754 Therefore, H03 is rejected, suggesting a significant nexus between MBS and the FP of licensed DMBs in Nigeria. Hypothesis four (H04) explores the influence of the Pay-direct system (PDS) on FP, yielding a beta coefficient of 0.336. The associated STDEV is 0.031, resulting in a T statistic of 10.683 and a p-value of 0.000. With a significance level of p<0.01, H04 is rejected, indicating a significant impact of PDS on the performance of license DMBs in Nigeria. Hypothesis five (H05) examines the relationship between remita (RMT) and FP, revealing a beta coefficient of 0.011. The STDEV associated with this coefficient is 0.040, yielding a T statistic of 0.289 and a p-value of 0.773, failing to reject the null hypothesis. This suggests that RMT does not significantly influence the financial performance of DMBs in Nigeria. Furthermore, the computation of the coefficient of determination (R2) is presented in Table 9. Table 9: Coefficient of Determination (R2). Construct R2 Performance 0.570 Source: SPSS Version 27, (2024). An R-square score of 0.570 signifies that roughly 57% of the variance in the dependent variable is elucidated by the independent variables incorporated in the model. In other words, factors such as Global Pay System and Corporate Internet Banking account for about 57% of the variability observed in the financial performance of licensed DMBs in Nigeria. Therefore, while the model provides valuable insights into the factors affecting DMBs' performance, it does not capture the entire complexity of the phenomenon. According to Chin's (1998) assertion, R2 values of 0.67, 0.33, and 0.19 should be regarded as substantial, moderate, and weak, respectively. The R2 value of 0.570, attributed to external latent variables affecting the target endogenous latent variable, is modest. 4.6 Assessment of Effect Size (f2) Effect size (f2) provides a measure of how strong the relationship is between variables in a statistical analysis. It goes beyond just determining if the relationship is statistically significant, helping researchers understand the practical importance of their findings (Ojeleye et al., 2023). A larger effect size indicates a stronger relationship between variables, while a smaller one suggests a weaker connection. By considering effect size alongside significance testing, researchers can better assess the real-world significance of their results and understand the impact of independent variables on the dependent variable (Haie et al., 2010). Essentially, effect size helps researchers gauge the actual strength of relationships in their data, making their interpretations more meaningful and applicable to real-life situations. Thus, the computation of the effect size is achieved using Cohen’s formula (Cohen, 1988; Hair et al., 2021) given as: Where: f 2 = the F-squared value that specifies the effect size of an exogenous variable's influence on an endogenous variable. R2Included = is the R2 value of the endogenous variable before omitting a particular exogenous construct. R2excluded = represents the changes in the R2 value of the endogenous variable after excluding a particular exogenous variable from a model. The f2 values of 0.02, 0.15, and 0.35 correspond to small, medium, and large effects, respectively (Cohen, 1988). In a manner akin to indirect relationship hypothesis testing, the alteration in R2 value upon the exclusion of a specific exogenous variable from the model serves to assess whether an omitted variable has a significant impact on the latent endogenous variable (Hair et al., 2014). Table 10 presents the results of the effect size analysis. Table 10: Assessment of Effect Size (f 2) Constructs FP Effect Size CIB 0.047 Small GPS 0.176 Medium MBS 0.010 Small PDS 0.220 Medium RMT 0.000 Nill Source: SEM-PLS Version 3.2.8 (2024). Table 10 assesses the effect size (f2) for various constructs in the analysis. The effect sizes indicate the magnitude of each construct's impact on the FP of license DMBs in Nigeria. CIB's effect size is 0.047, which is a small effect, suggesting a modest influence on