Multilevel relationships and loyalty in the microfinance industry: evidence from Tanzania

Purpose – The operations and viability of microfinance institutions (MFIs), crucial for socioeconomic developmentandpovertyreduction,heavilyrelyonthemultilevelrelationshipsamongborrowers,loanofficers andMFIs.Thisstudyexaminestherelationshipbetweeninterpersonalandfirm-levelrelationshipquality(RQ)andtheirsimultaneousimpactoncustomerloyalty(CL)inmicrofinance.Additionally,itinvestigatesthe mediatingeffectoffirm-levelRQbetweenCLandinterpersonalRQ. Design/methodology/approach – Inthisstudy, correlationalresearch methodswere employed. Completed questionnaires were received from 498 MFI borrowers in Dar es Salaam and Mwanza cities. Regression techniques and structural equation modeling were utilized to analyze the data. Before hypothesis testing, the validity and reliability of the measurements were confirmed. Findings – Interpersonal-level and firm-level RQs are significantly related. Interpersonal-level RQ and its dimensions are significantly linked to CL, whereas firm-level RQ and its dimensions are insignificantly related to CL, except for commitment. Interpersonal-level relationships have a stronger impact on CL than firm-level relationships. Among all the dimensions of RQ, commitment has the greatest influence on CL at both levels. Firm-level RQ negatively and insignificantly mediates the relation between interpersonal-level RQ and CL. Research limitations/implications – The study findings only apply to Tanzania ’ s microfinance industry, becausetheinteractionsbetweenandtherelativeeffectsoffirmandinterpersonaltiesmayvaryacrossvariouscontextsandcultures.Futureresearchmayconsiderreplicatingthisstudyinothercontextsandculturestoconfirmthesefindings. Practical


Introduction
In today's fiercely competitive global marketplace, achieving and maintaining customer loyalty (CL) has become paramount for securing a lasting competitive edge (Wijaya et al., 2022).CL refers to the tendency of customers who frequently acquire a good or service over time and hold positive opinions of either the product, service or the providing business (Tegambwage and Kasoga, 2023).The advantages of fostering CL are manifold, including increased repeat business, higher sales, business referrals, cost savings, positive word-ofmouth recommendations, references and enhanced publicity, all contributing to heightened profitability (Tegambwage and Kasoga, 2023).Thus, microfinance institutions (MFIs) need to cultivate and maintain a base of loyal customers to not only thrive and expand but also ensure long-term sustainability.This sustainability is of particular importance in low-income nations like Tanzania, where it plays a pivotal role in promoting financial inclusion, stimulating socioeconomic growth, and eradicating poverty (Tegambwage and Kasoga, 2022c).
In recent years, the financial services industry in Tanzania has achieved remarkable growth, intensifying competition among financial institutions and presenting challenges in building and retaining CL (Tegambwage and Kasoga, 2022a).An empirical study conducted by Kasoga and Tegambwage (2021) underscored the presence of multiple borrowings from different financial institutions and low levels of loyalty among Tanzanian micro borrowers, resulting in higher rates of over-indebtedness due to frequent switching behavior.The study revealed low switching barriers among financial consumers in Tanzania.Thus, MFIs in Tanzania face the challenge of enhancing and sustaining CL to ensure profitable growth and achieve their poverty reduction goals.In the financial services industry, long-term efforts to secure CL are not merely advantageous but a requirement to maintain sustained competitiveness (Tegambwage and Kasoga, 2022a).
One of the crucial factors in establishing CL is relationship quality (RQ) (Tegambwage and Kasoga, 2022a).Companies that have effectively cultivated quality relationships with their clients have achieved remarkable organizational benefits in terms of CL.Thus, scholars have emphasized the integration of relationship management into the CL development process (Tegambwage and Kasoga, 2023).This perspective aligns with social exchange theory (SET) (Blau, 1986), which posits that customers view loyalty building as a proactive, two-way opportunity rather than a passive, one-way connection, highlighting the importance of RQ in CL development.In the context of microfinance, the importance of quality relationships is amplified, as MFIs rely on these relationships to offer small loans to the poor who lack tangible collateral (Mbuya and Tegambwage, 2022).According to these scholars, a relational approach underpins all interactions between MFIs and borrowers, significantly influencing MFIs' capacity to collect loan repayments and meet stakeholder expectations.
Firm-customer relationships and their subsequent impact operate simultaneously at two levels: the interpersonal and firm levels (Palmatier et al., 2007).Nevertheless, few empirical studies have investigated the concurrent impacts of firm-and interpersonal-level interactions on firm performance and CL.For instance, Palmatier et al. (2007) discovered that buyersalesperson trust strongly influences a firm's financial performance compared to buyercompany trust within the manufacturing sector of the United States of America (USA).Similarly, Doney and Cannon (1997) discovered that both firm-and interpersonal-level trust influence purchasing intentions in an industrial manufacturing context in the USA.In the aviation and retail sectors in the USA, Sirdeshmukh et al. (2002) identified firm-level trust as more significant in the aviation sector and interpersonal trust as more critical in the retail sector.In the Australian retail landscape, Macintosh and Lockshin (1997) demonstrated that trust at the interpersonal level directly correlates with purchase intentions, whereas trust at the company level impacts purchase intentions through store attitudes.However, these findings exhibit variability across cultures and contexts suggesting that the associations and impacts of interpersonal and firm-level relationships vary.Thus, the findings of these prior empirical studies cannot be applied directly to the microfinance sector or low-income nations like Tanzania.In addition, these studies primarily focused on trust, one of the three fundamental components of RQ: trust, commitment and satisfaction (Tegambwage and JBSED Kasoga, 2022b).Furthermore, these previous studies did not explore the mediating role of firm-level RQ in the relationship between interpersonal-level RQ and CL.To the best of our knowledge, no study has concurrently investigated the effects of firm-and interpersonal-level RQ on CL.Consequently, understanding the simultaneous impact of RQ at both these levels on CL remains an unknown area.Notably, Palmatier et al. (2007) have advocated for further research to comprehend how multilevel relationships affect marketing outcomes such as CL.
This study employs the three fundamental RQ components of trust, commitment and satisfaction and investigates the role of firm-level RQ as a mediator in the relationship between interpersonal RQ and CL.Thus, the main objective of this investigation is to understand the interaction between interpersonal and firm-level RQ and the mechanisms through which they jointly influence CL, particularly within the understudied microfinance sector of a low-income country.Accordingly, we hypothesize that interpersonal-level RQ influences firm-level RQ and CL, firm-level RQ influences CL and firm-level RQ mediates the relationship between interpersonal-level RQ and CL.
This study carries several significant implications.By presenting a multilevel framework that elucidates how interpersonal-and firm-level RQ simultaneously affect CL, it contributes to the theoretical foundation of knowledge.Moreover, it highlights the role of firm-level RQ as a mediator in the link between interpersonal-level RQ and CL.From a practical standpoint, these findings improve diagnostics for understanding, improving and sustaining CL from a relational perspective.More specifically, this study furnishes policymakers and microfinance service providers with valuable insights into the importance of enhancing interpersonal-level RQ as a strategy to enhance firm-level RQ and maintain CL-a critical element for the long-term sustainability of MFIs.In turn, MFIs' sustainability in low-income countries like Tanzania is vital for stimulating socioeconomic development and, ultimately, the achievement of the poverty eradication goals (Tegambwage and Kasoga, 2022c).

Literature review Theoretical review
The social judgment theory (SJT) (Hamilton and Sherman, 1996) and SET (Blau, 1986) are two theories that provide support for the relationships outlined in the model proposed (Figure 1).SJT claims that individuals evaluate both individuals and groups using various techniques.According to Palmatier et al. (2007), strong and confident opinions are formed about people based on previous deductions, as dispositional qualities are believed to lead to similar future behaviors.Contradictory information is commonly ignored or attributed to situational factors because existing views serve as anchors (Hamilton and Sherman, 1996).By contrast, when analyzing a company, people tend to form weaker and slower judgments (Hamilton and Sherman, 1996).Therefore, judgments about an individual have a greater impact on the client's attitudes and actions than judgments about a corporation (Palmatier et al., 2007).This

Multilevel relationships
theory is relevant for explaining how borrowers evaluate their relationships with both the loan officer and the MFI.In particular, the theory explains why borrowers might assess these relationships differently.
On the other hand, the association between RQ at both the interpersonal and firm levels and CL can be explained by SET.According to this theory, all interactions among people are determined by weighing the costs and benefits of various options (Blau, 1986).For instance, if someone perceives that leaving a relationship is more costly than remaining in it, or vice versa, they will make their decision accordingly.SET posits that those who give a lot to others seek to gain a lot in return, while those who receive a lot from others feel obligated to reciprocate.In other words, the perception of being well-treated by one party fosters a sense of obligation to repay the other party.Consequently, social exchange relationships develop through a sequence of reciprocal transactions between individuals or parties, creating a pattern of mutual obligations.Accordingly, relationship commitment among individuals depends on the ongoing comparison of social and economic outcomes across a series of interactions with other parties and available alternatives (Blau, 1986).SET is well-suited for this study because interactions between borrowers and the MFI through loan officers can be viewed as social exchanges that lead to borrowers' loyalty to the MFI.

Empirical review
As mentioned earlier, no study has explored how multilevel interactions affect CL in the setting of microfinance, especially in low-income nations like Tanzania.However, a handful of empirical studies have investigated the concurrent impacts of firm-and interpersonal-level interactions on firm performance and CL.For example, Palmatier et al. (2007) investigated how customer interactions with the company and its salespeople impact a company's financial results in the manufacturing sector of the USA.They discovered that the buyersalesperson RQ strongly influences the firm's financial results compared to the buyercompany RQ.Doney and Cannon (1997) explored how trust at both the firm and interpersonal levels affects purchase intentions in an industrial manufacturing scenario in the USA.They found that trust at both levels is associated with purchasing intentions.Sirdeshmukh et al. (2002) explored the effects of firm-level and interpersonal-level trust on CL in the aviation and retail sectors in the USA.They found that trust has opposite effects on CL depending on the level of trust, with firm-level trust being more significant in the aviation environment and interpersonal-level trust being more crucial in the retail sector.Macintosh and Lockshin (1997) studied the effects of firm-and interpersonal-level trust on consumers' purchase intentions in the context of Australian retail.They discovered that trust at the interpersonal level is directly related to purchase intentions, while trust at the company level is linked to purchase intentions via shop attitudes.
Although these empirical studies reported mixed findings, suggesting that the associations and impacts of interpersonal-and firm-level relationships vary across various cultures and contexts, their strength lies in their findings that interpersonal-and firm-level relationships have different impacts on financial performance, purchase intentions and CL.Their limitations, however, are that they primarily focused on developed countries and other industries, rather than microfinance and concentrated on one dimension of RQ-trustwhile ignoring other important dimensions of RQ such as commitment and satisfaction.In addition, the aforementioned empirical studies did not investigate the mediating role of firmlevel relationships.Nevertheless, since trust is a component of RQ (Tegambwage and Kasoga, 2023), these empirical studies provide valuable insights into the potential impact of firm-and interpersonal-level RQ on CL in the present study.

Hypotheses development
According to Hennig-Thurau et al. (2002), RQ is a term used to describe the degree to which a relationship can fulfill the needs of the consumer.The client's relationship with a salesman can influence the client's bond with the selling firm (Sirdeshmukh et al., 2002).A stronger client-salesperson RQ is likely to increase the client-company RQ, provided that the salesperson is associated with that company (Sirdeshmukh et al., 2002), as the salesman is seen as a representative of the seller.In line with this, Wijaya et al. (2022) found that customer-salesperson trust positively and significantly impacts company trust.Similarly, Sirdeshmukh et al. (2002) highlighted the asymmetrical connection between customer trust in salespeople and organizations, suggesting that if a salesperson is not trustworthy, the customer may not trust the company.MFIs build relationships with borrowers through routine interactions between loan officers and borrowers, which helps reduce monitoring, defaulting and collection costs (Tegambwage and Kasoga, 2022c).Therefore, it is reasonable to assume that borrower-loan officer RQ at the interpersonal level may influence the quality of relationships that borrowers have with the MFI at the firm-level RQ.Thus, it is postulated that: H1. Interpersonal-level RQ is significantly and positively linked to firm-level RQ.
According to Wu et al. (2019), CL refers to a customer's commitment to using a financial institution's service for a predetermined period.Maintaining solid bonds with customers fosters loyalty benefits, such as their willingness to recommend, pay more, spend more and purchase more (Tegambwage and Kasoga, 2022a).Empirical research has consistently documented a strong and favorable impact of interpersonal-level interactions on CL (Doney and Cannon, 1997;Macintosh and Lockshin, 1997).For instance, Doney and Cannon (1997) reported the positive and significant direct effects of interpersonal-level interactions on purchase intentions in industrial marketing.Interpersonal-level trust exerts a more substantial impact on CL in the retail context compared to the airline context (Sirdeshmukh et al., 2002), suggesting that contextual variations alter the connection between interpersonal-level interactions and CL.In the microfinance sector, which relies on borrower-loan officer relationships, an improvement in the quality of a borrower's relationship with the loan officer is expected to result in increased loyalty to the MFI, based on SET and empirical research.Accordingly, the individual dimensions of interpersonal-level RQ, namely commitment, trust and satisfaction, are expected to influence CL in the microfinance industry.Thus, it is proposed that: H2. Interpersonal-level RQ is significantly and positively linked to CL.
H2a. Commitment to the loan officer is significantly and positively linked to CL.
H2b. Trust in the loan officer is significantly and positively linked to CL.
H2c. Satisfaction with the loan officer is significantly and positively linked to CL.
Empirical research has also demonstrated the positive effects of firm-level relationships on CL (Tegambwage and Kasoga, 2022a).For example, Doney and Cannon (1997) reported the direct positive effects of firm-level interactions on purchase intentions in industrial marketing.Macintosh and Lockshin (1997) found that firm-level trust is indirectly related to purchase intention through shop attitudes in the retail environment.However, Sirdeshmukh et al. (2002) discovered that firm-level trust has a greater influence on CL in the airline sector compared to the retail setting, suggesting that contextual variations influence the effects of firm-level interactions on CL.In the microfinance sector, where the quality of a borrower's relationship with an MFI is important, it is expected that an improvement in this relationship will lead to increased loyalty to the MFI based on SET and empirical research.Accordingly, each of the three dimensions of firm-level RQ, namely, commitment, trust and satisfaction, is expected to influence CL in the microfinance context.Thus, it is proposed that: H3. Firm-level RQ is significantly and positively linked to CL.
H3a. Commitment to MFI is significantly and positively linked to CL.
H3b. Trust in MFI is significantly and positively linked to CL.
H3c. Satisfaction with MFI is significantly and positively linked to CL.
Previous studies have demonstrated a significant positive relationship between firm-level RQ and CL (Tegambwage and Kasoga, 2022a) as well as between interpersonal-level RQ and CL (Doney and Cannon, 1997).Additionally, as mentioned earlier, prior investigations have found a strong correlation between interpersonal and firm-level interactions (Palmatier et al., 2007;Sirdeshmukh et al., 2002).Since loan officers manage MFI-borrower relationships through frequent meetings with borrowers (Tegambwage and Kasoga, 2022b), it is expected that loan officer-borrower RQ will influence MFI-borrower RQ.Thus, since interpersonal relationships influence firm-level relationships, and firm-level relationships influence CL, it is expected that firm-level RQ might mediate the relationship between interpersonal-level RQ and CL.Therefore, it is proposed that: H4. Firm-level RQ mediates the relationship between interpersonal-level RQ and CL.
Figure 1 displays a model illustrating the connection between multilevel RQ and CL.

Methodology
This investigation employs a correlational technique, with the microfinance industry serving as the ideal testing ground for our theoretical framework.In microfinance services, the presence of borrower-loan officers and borrower-MFI relationships makes it easy to distinguish between the effects of these relationships on the CL.Borrowers establish strong bonds with both loan officers and MFIs through several interactions.Therefore, the study population consisted of all MFI borrowers from Dar es Salaam and Mwanza cities where a significant number of Tanzania borrowers are located (Kasoga, 2020).A systematic sample of 900 borrowers was administered questionnaires as they exited various MFIs in the two cities using a step-of-three approach.Systematic sampling was chosen instead of simple random sampling due to the lack of a sampling frame (Hair et al., 2019).MFIs were reluctant to provide the author with a list of borrowers due to confidentiality concerns.To ensure voluntary and honest participation, the survey's purpose was explained, and confidentiality was assured to respondents (Owusu et al., 2021).There were 498 viable responses in total, with a 55.3% response rate across exits.The majority of respondents were female (70.8%), in line with Kasoga and Tegambwage's (2021) findings, as women make up the vast majority of MFI borrowers in Tanzania.Respondents' ages ranged from 18 to 49 years, with the majority (47.0%) falling between 26 and 35 years.Most of the respondents (64.5%) were married and the majority (79.1%) had completed their primary education.
The data collection tool's validity and reliability were assessed using factor analysis (Anderson and Gerbing, 1988).Keiser-Meyer-Olkin (KMO) and Bartlett tests were conducted to determine sample sufficiency for factor analysis, with a KMO result of 0.872 (>0.5) and a significant Bartlett's test of sphericity (p < 0.01) confirming the adequacy of the sample size (Hair et al., 2019).Reliability was assessed using Cronbach's alpha coefficients (α): 0.910 for interpersonal-level RQ, 0.910 for firm-level RQ and 0.812 for CL.These coefficients exceeded the recommended criterion of 0.7, and factor loadings with weights greater than 0.5 (ranging JBSED from 0.519 to 0.997) indicated high reliability (Hair et al., 2019).Convergent validity was established as factor loadings were statistically significant (p < 0.01) (Anderson and Gerbing, 1988), and discriminant validity was confirmed as the square root of the average variance extracted (AVE) exceeded the correlations between the variables (Table 2) (Hair et al., 2019).Common method variance was minimized by using existing measures and guaranteeing respondents' anonymity (Field, 2009).The level of multicollinearity among the explanatory measurement variable was assessed through a multiple regression analysis, calculating the variance inflation factor (VIF).VIF results in Table 3 ranged from 1.008 to 1.873 (<5), indicating that the factors were not substantially correlated (Hair et al., 2019).
Research variables operationalization was based on items validated in previous studies.Measures of commitment, trust and satisfaction were derived from Tegambwage and Kasoga (2022c) and were also used to calculate a composite score for the RQ construct (Tegambwage and Kasoga, 2022c).Two items adapted from Tegambwage and Kasoga (2023) were used to gauge the CL, employing a 5-point Likert scale with the options ranging from (1) strongly disagree to (5) strongly agree for each item.These items were modified for the microfinance context and validated through a pretest involving ten microfinance specialists.The revised questionnaire was further tested with 20 different borrowers in Dodoma City, Tanzania, before its approval as the final version.

Research findings
To summarize the observed data, means and standard deviations (SD) were computed (Table 1).The mean score of 2.02 for the CL construct indicates that borrowers in Tanzania are often not loyal to MFIs, most likely due to their ability to easily switch between MFIs (Kasoga and Tegambwage, 2021).The borrower's RQ with both the loan officer and the MFI is low in Tanzania, with mean scores of 1.89 and 2.56, respectively.The mean scores for all RQ dimensions, except trust in the MFI, range from 1.37 to 2.01, indicating weak interpersonal and firm-level relationships.However, borrowers' trust in MFIs is high, with a mean score of 4.08, indicating that Tanzanian borrowers have greater faith in MFIs.The low standard deviations relative to their mean values suggest that statistical means are a good fit for the observed data (Field, 2009).The skewness and kurtosis values in Table 1 fall within the acceptable range (Hair et al., 2019), indicating a normal distribution.
To determine whether there were any linear correlations between the constructs, a Pearson correlation analysis was conducted (Field, 2009), and the findings are presented in Table 2.According to the findings, interpersonal and firm-level RQ are significantly correlated in a positive direction (r 5 0.35, p < 0.01), suggesting that as the borrower's RQ

Multilevel relationships
with the loan officer improves, so does the borrower's RQ with the MFI.This finding supports H1, which posits that interpersonal-level RQ is positively related to firm-level RQ.The findings also indicate a strong and positive correlation between interpersonal-level RQ and CL (r 5 0.48, p < 0.01), supporting H2 which proposes that RQ is positively related to CL at the interpersonal level.This conclusion implies that an improvement in the borrower's relationship with the loan officer increases their levels of loyalty to the MFI.Furthermore, the findings show that firm-level RQ is insignificantly linked to CL (r 5 0.05, p > 0.05), rejecting H3.

JBSED
At the dimensional level, all correlations with CL are positive and significant, except for trust and satisfaction with the MFI, which are negative and insignificant.This suggests that H2a, H2b, H2c and H3a are supported, whereas H3b and H3c are rejected.However, it should be noted that these results of the correlation analysis offer the first indication of whether the study's hypotheses are validated.Hence, an additional analysis (regression analysis) was conducted to confirm these hypotheses.
Multiple regression analysis was employed to assess the predictive power of the predictor factors on the criterion variables (Hayes, 2017).The results in Table 3 reveal that interpersonal-level RQ positively and significantly affect firm-level RQ (β 5 0.408, p < 0.001), accepting H1.Furthermore, interpersonal-level RQ exhibits a positive and significant influence on CL (β 5 0.897, p < 0.001), suggesting that as borrower-loan officer RQ improves, so does their loyalty to the MFI, supporting H2.Conversely, firm-level RQ has a negative and insignificant impact on CL (β 5 À0.196, p > 0.05), rejecting H3.
Additionally, when examined at the dimensional level, Table 3 indicates that all relationships between interpersonal-level relationships and CL are positive and significant: commitment (β 5 0.460, p < 0.001), trust (β 5 0.403, p < 0.001) and satisfaction (β 5 0.292, p < 0.001), supporting H2a, H2b and H2c.On the other hand, the connections between firmlevel relationships and CL at the dimensional level are as follows: commitment (β 5 0.218, p < 0.001), trust (β 5 0.002, p > 0.05) and satisfaction (β 5 0.048, p > 0.05), confirming H3a and rejecting H3b and H3c.It is worth noting that all interpersonal-level relationships exert a stronger impact than their firm-level counterparts, with commitment having the most significant influence, followed by trust and satisfaction, in that order.

Robustness checks
Robustness checks were conducted to validate the findings.Specifically, a regression analysis was employed to determine the predictive power of interpersonal-and firm-level RQ on CL while considering the respondent's socio-demographic factors (age, gender, education status and marital status) as control variables.The analysis shows that age (β 5 0.031, p > 0.05), gender (β 5 0.041, p > 0.05), educational status (β 5 0.012, p > 0.05) and marital status (β 5 0.009, p > 0.05) do not have significant effects on CL (Table 3).This suggests that the impact of interpersonal-level and firm-level RQ on CL remains robust, regardless of these sociodemographic factors.

Discussion
The findings of this study corroborate H1 by demonstrating a positive and significant relationship between interpersonal-level and firm-level RQ.Specifically, the study reveals that interpersonal-level RQ has a positive and significant effect on firm-level RQ.This suggests that a stronger borrower-loan officer relationship translates into a more favorable MFI-borrower relationship.These findings align with prior studies (Palmatier et al., 2007;Sirdeshmukh et al., 2002) and can be attributed to the role of loan officers in microfinance services.Loan officers frequently engage with borrowers through frequent meetings (Tegambwage and Kasoga, 2022b), managing and nurturing relationships, which, in turn, shape the borrower's perception of the loan officer as an agent of the MFI.Strong borrower RQ contributes to an enhanced borrower-MFI RQ because borrowers perceive the loan officer as an agent of the MFI.
Additionally, the findings indicate that at the interpersonal level, RQ has a positive and significant impact on CL, supporting H2.In addition, all RQ dimensions at the interpersonal level exhibit positive and significant effects on CL, supporting H2a, H2b and H2c.Hence, MFIs should consider implementing innovative strategies aimed at enhancing the borrowerloan officer RQ to foster CL.For instance, MFIs must ensure that loan officers gain the commitment and trust of borrowers by providing training in building and maintaining strong relationships with borrowers.Borrowers expect loan officers to be professional, friendly, trustworthy, responsive and empathetic.This result is in line with previous research (Macintosh and Lockshin, 1997;Sirdeshmukh et al., 2002).Hence, MFIs should ensure that loan officers are appropriately trained and equipped with adequate resources to create unique and positive experiences for borrowers, ultimately increasing CL.
Furthermore, the results reveal that firm-level RQ has a negative and insignificant impact on CL, thus rejecting H3.This finding contrasts with Tegambwage and Kasoga (2022a).Additionally, the findings indicate that all RQ dimensions at the firm-level have positive, but insignificant effects on CL, except for commitment, which has a positive and significant effect on CL.Hence, H3a is accepted, whereas H3b and H3c are rejected.This suggests that at the firm-level, the effects of trust and satisfaction on CL may be transmitted through commitment.
The results demonstrate that the two RQ levels have different effects on loyalty, with interpersonal relationships exerting a stronger influence on CL than firm-level relationships.This aligns with SJT.Likewise, Palmatier et al. (2007) pointed out that relationships with individuals have a greater impact on outcomes than those with businesses.It is noteworthy that commitment, among all the dimensions of RQ at both the interpersonal and firm levels, has the greatest impact on CL.Thus, MFIs should focus on fostering the borrower's commitment to both the loan officer and MFI to enhance CL.This can be achieved by educating and empowering loan officers to promote satisfaction and trust in all interactions with borrowers.According to Hennig-Thurau et al. (2002), commitment cannot be built without trust and satisfaction.

JBSED
Regarding the mediation effect, a negative and insignificant indirect effect of interpersonal-level RQ on CL through firm-level RQ is revealed.This means that firm-level RQ does not mediate the effect of interpersonal-level RQ on CL, thus rejecting H4.This finding implies that the effect of interpersonal-level RQ on CL is not transmitted through firm-level RQ.This is a unique finding because, as far as we are aware, no study has reported the mediation effect of firm-level RQ between interpersonal-level RQ and CL.Therefore, the most effective strategy to enhance CL in the microfinance industry is to improve borrower-loan officer RQ.Indeed, strong borrower-loan officer relationships are advantageous as they foster a positive flow of goodwill toward the borrower-MFI relationship and increase CL towards MFI.In turn, higher levels of CL will ensure the sustainability of MFIs (Tegambwage and Kasoga, 2022c), ultimately contributing to poverty reduction goals.MFIs can build and maintain high-quality interpersonal relationships between loan officers and borrowers by providing training and empowerment to loan officers to foster commitment, trust and satisfaction in all interactions with borrowers.

Conclusion
The empirical findings reveal a positive and significant relationship between interpersonal-level RQ and firm-level RQ.However, CL is impacted differently by these two levels of RQ.At the interpersonal level, RQ and its dimensions, namely commitment, trust and satisfaction, exhibit a positive and significant relationship with CL.In contrast, at the firm level, RQ and its dimensions, namely trust and satisfaction, show a positive but not significant relationship with CL.Only the commitment dimension is positively and significantly related to CL at the firm level.Importantly, the interpersonal-level RQ and its dimensions exert stronger effects on CL than the firm-level RQ and its dimensions.Among all the dimensions of RQ at both interpersonal and firm levels, commitment has the most significant impact on CL.Furthermore, firm-level RQ negatively and insignificantly mediates the relationship between interpersonal-level RQ and CL.
These results have significant theoretical and practical ramifications.From a theoretical perspective, this study contributes to the literature by proposing a multilevel framework to elucidate how RQ at interpersonal and firm levels concurrently influences CL.This framework provides a more thorough understanding of how interpersonal-level RQ affects firm-level RQ and highlights the mediating role of firm-level RQ in the relationship between interpersonal-level RQ and CL-a novel finding in the field.From a practical standpoint, this study advances our knowledge of CL and the role of RQ at both interpersonal and firm levels in establishing CL.This understanding is particularly relevant for the sustainability of MFIs in low-income nations like Tanzania.Sustainable microfinance services are essential for fostering financial inclusion, driving socioeconomic growth and ultimately alleviating poverty in such regions.To this end, authorities and MFIs' management should prioritize the development of commitment, trust and satisfaction in all interactions between borrowers and loan officers, notably through adequate training and empowerment of loan officers.
However, it is important to note that this study, while comprehensive and providing a framework of the factors that influence CL from a relational perspective, is based on a sample of 498 respondents, suggesting room for further research.Future studies could employ larger, nationally representative samples to enhance the generalizability of the model.Second, the contextual and cultural variations may limit the applicability of these findings beyond Tanzania's microfinance industry.Replicating this study in different contexts and cultures can help validate these results.Third, given that this study is cross-sectional, it does not account for behavioral changes over time.Longitudinal designs should be considered in future research.Finally, while this study focuses on commitment, trust and satisfaction as criteria for a good relationship, future studies may explore other dimensions of the RQ construct, such as customer orientation, expertise, opportunism, cooperative norms and conflict resolution.

Multilevel relationships
Figure by author Figure 1.Proposed research model

Table 1 .
LO stands for loan officer.MFI stands for microfinance institution.RQ stands for relationship quality Source(s): Table by author Descriptive statistics Adjusted R 2 (CL) 5 0.564; F 5 108.354***;AdjustedR 2 (Firm-level RQ) 5 0.122; F 5 70.309***CL stands for customer loyalty.LO stands for loan officer.MFI stands for microfinance institution.RQ stands for relationship quality.***p< 0.001 Source(s):Table by author Square root of AVE listed on diagonal.The off-diagonal elements are correlations between the constructs.LO stands for loan officer.MFI stands for microfinance institution.RQ stands for relationship quality.**p < 0.01.*p < 0.05 Source(s): Table by author

Table 4 .
Mediation analysis results