Bank choice behavior, customer participation, a study in the retail banking sector

search,experience and credence influence customer loyalty; b and through which mechanismcustomer participation influence the relationships between attributes of bank choice behaviorand c

At the first, I would like to thank all those people who made this thesis possible and anunforgettable experience for my studying Foremost, I would like to express my sinceregratitude to my supervisor, Dr Ngo Viet Liem, for his intensive support, valuable suggestions,instructions and encouragement during the time of doing my research His guidance helped me

in all the time of research and writing of this thesis

Besides my supervisor, I would like to thank the rest of my thesis committee: Prof.Nguyen Dong Phong, Prof Nguyen Dinh Tho, and Dr Tran Ha Minh Quan … for theirencouragement, insightful comments, and hard questions

I thank my classmates in ISB MBUS 2011 (specially members in my group: Ms Hanh,

Ms Hang, Mr Chuong and Ms Hien) and my colleagues at Shinhan Bank for their encouragingand supporting to me complete this thesis Besides that, my sincere thanks are given to all of myteachers at International Business School – University of Economics Ho Chi Minh City for theirteaching and guidance during my MBUS course

Last but not the least, I would like to thank my family I must express my gratitude toPhan Thi Hoi, my older sister, for her continued support and encouragement I also wish thankall those people who spent through their time and generous support made this thesis project

Ho Chi Minh City, October, 2014

PHAN DANH HOANG

This study integrates bank choice behavior and customer participation into a theoreticalframework to explain variation in customer loyalty Specifically, the purpose of this study istwofold: (a) to examine the extent to which attributes of bank choice behavior (e.g search,experience and credence) influence customer loyalty; (b) and through which mechanismcustomer participation influence the relationships between attributes of bank choice behaviorand customer loyalty The empirical setting of this study is the retail banking sector A survey of

466 customers using personal loan services in some domestic banks is conducted

The findings indicate that bank choice behavior and customer participation playimportant role in predicting the loyal of customer who use personal loan services The resultsalso support that customer participation is biggest impact to customer loyalty, followed bycredence attributes and experience attributes Whereby, the banking managers need to considercreating all of experience attributes, credence attributes and customer participation to make theircustomer loyalty to the bank The research results are limited in the power of measurementscale, sampling approach as well as the fitness of the research model and data It results in thevaluable directions for further researches in future

Keywords

Retail banking, personal loan, bank choice behavior, customer participation, customer loyalty

TABLE OF CONTENTS

ACKNOWLEDGEMENT 1

ABSTRACT 2

TABLE OF CONTENTS 3

CHAPTER 1 - INTRODUCTION 7

1.1 Background to the research 7

1.2 Statement of the problem 8

1.3 Research questions 8

1.4 Significance of the Study 9

1.5 Research scope and delimitation 9

1.6 Organization of the thesis 10

CHAPTER 2 11

THEORETICAL MODEL AND HYPOTHESES 11

2.1 Bank choice behavior 11

2.2 Customer participation: 12

2.3 Customer Loyalty 15

2.4 Research model and hypotheses development 16

Building upon prior research, Figure 2.2 integrates the focal constructs of the thesis with hypothesized relationships 16

2.4.1 Bank choice behavior and customer loyalty 17

2.4.2 Bank choice behavior and customer participation 18

2.4.3 Customer participation and customer loyalty 20

CHAPTER 3 22

RESEARCH METHODS 22

3.1 Research design 22

3.2 Development of questionnaire 23

3.2.1 Measurement scales 23

3.2.1.1 Measure of search, experience and credence 23

3.2.1.2 Measure of customer participation 25

3.2.1.3 Measure of customer loyalty 26

3.2.2 Draft questionnaire 26

3.3 Pilot study 27

3.4 Sample method 28

3.5 Questionnaire administration 30

3.6 Data analysis methods 31

CHAPTER 4 32

DATA ANALYSIS AND FINDINGS 32

4.1 Descriptive analysis 32

4.2 Reliability analysis 33

4.3 Exploratory Factor Analysis (EFA) 36

4.4 Confirmatory Factor Analysis 39

4.4.1 Composite reliability and variance extracted 41

4.4.2 Unidimensional analysis 42

4.4.3 Convergent validity analysis 42

4.4.4 Discriminant validity analysis 43

4.5 Research model test 43

4.5.1 Theoretical model test by using SEM approach 43

4.5.2 Theoretical model estimation by BOOTSTRAP 46

CHAPTER 5 48

DISCUSSION, IMPLICATIONS AND LIMITATION 48

5.1 Discussion 48

5.2 Implications of the research 49

5.3 Limitations and directions for further research 51

REFERENCES 53

APPENDICES 61

APPENDICES APPENDIX 1 - Questionnaire 61

APPENDIX 2 - The results of population census 2009 in Ho Chi Minh City 67

APPENDIX 3 - The result of EFA analysis – Eigenvalues for variables (the first test) 67

APPENDIX 4 - CFA test result 72

LIST OF FIGURES Figure 2.1 - Totally disaggregated second-order model of bank choice behavior 12

Figure 2.2 - The proposed research model with hypotheses 21

Figure 3.1 - Research process 23

Figure 4.1 - CFA result in the second time (Standardized estimates) 40

Figure 4.2 - Result of hypothesis test in the second time 45

Figure 4.3 - The result of CFA in the first time (Standardized) 72

LIST OF TABLES Table 3.1 - Measurement scales of bank choice behavior 24

Table 3.2 - Measurement scales of customer participation 25

Table 3.3 - Measurement scales of customer loyalty 26

Table 3.4 - calculating sample with quota sampling method 29

Table 4.1 - Descriptive statistic of respondent’s characteristics 32

Table 4.2 - Descriptive statistic of respondent’s characteristics 33

Table 4.3 - Cronbach’s alpha reliability coefficient 34

Table 4.4 - Cronbach’s Alpha coefficients for each measurement scale 34

Table 4.5 - The EFA analysis result 37

Table 4.6 - Cronbach’ Alpha results after excluding unsatisfactory items 38

Table 4.7 - Modification indices 39

Table 4.9 - Discriminant validity of concept 43

Table 4.10 - Regression Weights 44

Table 4.11 - Regression Weights: (Group number 1 - Default model) 45

Table 4.12 - The results for Bootstrap estimation 47

CHAPTER 1 - INTRODUCTION 1.1 Background to the research

The Vietnamese banking sector is transforming at a rapid pace and is expected to show ahigher growth rate than most of its counterparts in neighboring Asian countries in the upcomingyears (Ho and Baxter, 2011) This tendency has been mainly shaped by state-headed economicreconstructions since the end of the 1980s and the creation of several State-Owned CommercialBanks (SOCB) It has accelerated recently with the country’s accession to the World TradeOrganization in 2007, heading to a dynamic liberalization of the banking sector These reformsand the tremendous economic development of the country during the last two decades (Breu andDobbs, 2012) are making important chances for both national and foreign banks operating inVietnam, mainly in retail banking sector

Since early 2014, although interest rates for enterprises were fall down in the lowest levelsince the stimulus package in 2009 of Government, but the credit growth is still difficult toincrease Many enterprises still hesitate to get loans So many banks shift their emphasis topersonal loans services to grow credit outstanding balance In particular, personal loan servicesattracted attention from many individual customers by simple and convenient procedures Thesuccess of retail banking sector depends on the type of relationship maintained with its customerand an understanding of customer needs In the commercial banking industry there is a growingrecognition that individual customers not only represent a viable market segment but that theirneeds are different and frequently changed

One strategic tool that banks can use to attract new customers, retain current customersand develop strong relationships is the customer loyalty Customer loyalty is one of keyconcerns for researchers and marketers who are interested in understanding its predictors andcausal structure Previous research on this concept also showed that customer participationleading to loyalty (Auh, Bell, McLeod& Shih, 2007; Ngo &O'Cass, 2013) It gives the notionthat customer participation is very important concept for marketers and managers

Besides the customer loyalty, bank choice behavior attributes (search, experience,credence) also attract many of researchers (Mitra, Reiss, and Capella, 1999; Babakus,Eroglu&Yavas (2004) However, few empirical studies investigate the relationship between bankchoice behavior contributes, customer participation and customer loyalty This study thereforedevelops a model examining the role of bank choice behavior, customer participation oncustomer loyalty Empirically, it tests the impact of bank choice behavior attributes on customerparticipation Then, it investigates the impact of bank choice behavior contributes and customerparticipation on customer loyalty Following this introduction, the next section presentsstatement of the problem.

1.2 Statement of the problem

The demand for personal loan products has been increasing dramatically However, banksstill do not have an optimal recipe for attracting new customers, retaining and selling moreservices to existing customers A study on the effects of bank choice behavior and customerparticipation on customer loyalty will allow bank managers to have an appropriate marketingstrategy to attract new customers It can help bank managers give out methods which help toincreases the level of customer loyalty, thereby maintaining the existing customers Throughstudying the relationship between customer participation and customer loyalty, it also allowsbank managers find out ways to strengthen the participation of the customer in the transactionprocess through which banks can increase the level of customer loyalty

1.3 Research questions

The purpose of this research is to empirically investigate the impact of bank choicebehavior and customer participation on customer loyalty The empirical setting in this particularresearch is the retail banking sector in the transitional economy in Vietnam Contrary to the past,many customers are trying to use personal banking services and have a lot of products to choose.Given this situation, a study of bank choice behavior, customer

participation and customer loyalty would be very useful This study attempts to answer the following research questions.

 To what extent, do attributes of bank choice behavior (e.g search, experience and credence) and customer participation influence customer loyalty?

 Between bank choice behavior and customer participation, which factor will most influence to customer loyalty?

1.4 Significance of the Study

This study has important implications for both practitioners (leaders of commercialbanks, marketing managers) and academic (researchers, students of the business administrationdepartment) as follow:

 The findings of this study contribute to research on bank choice behavior andcustomer participation by investigating the contribution of both bank choice behavior andcustomer participation to enhancing customer loyalty Based on it, managers ofcommercial banks will make an effective strategy to enhance customer loyalty andimprove operational efficiency

 Marketing managers will consider the factors of bank choice behavior whichaffect customer loyalty in personal loan market sector to set up an appropriatecommunicate strategy The results of this study also are a basis for researchers, studentsfor developing further research to practical applications

1.5 Research scope and delimitation

This study is conducted in Ho Chi Minh City, one of the biggest economic centers of VietNam Respondents are customers who are using personal loan services at banks Due to limitedtime and cost, the research surveyed about 500 customers of large domestic commercial banks inHCMC The result of research in this city, in some level can represent for Viet Nam in generaland can be use as reference for further purposes

The research will assess the effect of bank choice behavior attributes and customerparticipation to customer loyalty In this, the quota sampling method is used to collect researchdata It is a type of the non-probability sampling Althought selected-sample is not generalized topopulation, if we choose the right controlled characteristic (distinguishing attributes of studiedobjects), in theoretically, sample cannot represent for population but in practice it may representfor population.

1.6 Organization of the thesis

This thesis is organized in five chapters The first chapter is the introduction thatdescribes the overview of research background, research problem, and objective, followed bythe scope of research, implications, and structure of thesis

Chapter 2 reviews previous research on customer loyalty and customer participation.Specifically, the chapter reviews the mediating role of customer participation model and theory

of bank choice behavior and, how they are used in various contexts This chapter concentrates

on explaining each variable in the model, and reasons for choosing them to be included in theresearch model

Chapter 3 introduces research methodology used to validate the research model inprevious session It presents the research design, development of survey questionnaire,qualitative study, and main survey This chapter also presents how to collect data and analyze thedata collected to test the research hypotheses proposed in chapter 2

Chapter 4 analyses data as well as discusses the result finding in connection with researchmodel This chapter explains the empirical part of the study This part discusses the analyses ofthe data received, psychometric properties of focal constructs, and hypothesis testing

The last chapter, chapter 5 discusses implications and research limitations Finally, thisthesis makes suggestions for further research on the topic area

References and appendixes are included in the end of thesis

CHAPTER 2 THEORETICAL MODEL AND HYPOTHESES

This chapter provides a theoretical background about bank choice behaviour, customerparticipation, customer loyalty, upon which a conceptual research model and hypotheses areconstructed

2.1 Bank choice behavior

The underlying theory of the bank choice behavior model was first developed within theeconomics of information literature (Darby and Karni, 1973; Nelson, 1970) and later refinedwithin the services context (Bloom and Pailin, 1995; Lynch and Schuler, 1990) Three sets ofattributes are hypothesized to be manifestations of a consumer’s choice decisions: search,experience and credence (Mitra et al., 1999) as shown in Figure 1 Search attributes are thosethat are accurately evaluated before making a choice decision (Srinivasan and Till, 2002)According to Babakus at al., (2004), search, experience and credence can defined as follows:

 Search attributes – which may be accurately evaluated before making a choice.Within the banking context, these include interest rates on deposits or loans, overdraftprivileges, fees and charges, working hours and new bank services

 Experience attributes – which may be accurately assessed only when the servicehas been purchased and used Within the banking context, these include the helpfulness oftellers, the supply of services with a timely manner, the attentiveness of bank employees,their temperament to pay attention to the client, their friendliness and expertise, and thespeed of decision made by the bank authorized person

 Credence attributes – which can or might not be evaluated even when a sale iscreated, simply because the client might lack the technical experience needed to create theassessment or there may be without delay accessible indicators of the attributes Within thebanking context, credence attributes consist keeping the client’s banking information

confidential, the integrity of the bank, and adoption of bank in receiving of the most recent technology, its management and its overall ability inside the business of banking.

Figure 2.1 - Totally disaggregated second-order model of bank choice behavior

Most managers grasp that paying attention to customers makes sensible business sense Bankshave abundant to achieve from actively seeking and inspiring client participation, that bankshave a tendency to outline as obtaining customers to supply constructive suggestions and sharetheir concepts on the way to form product and repair offerings Nonetheless whereas the thoughtthat soliciting and listening rigorously to customers is recent, several corporations solely pay lipservice to it The return on customer participation is higher than several managers suppose.

Dabholkar (1990) considered customer participation as involved level of customers in theproduction and supply of services Integrating customers into the innovation process ofcompanies will help companies improve and enhance the quality of service and success in themarket Dagger & Sweeney (2007) also claimed that improving service quality in the serviceindustry is the top priority because high quality services not only increases the positive intention

of customers behavior but also contribute to improved market share, revenue and greaterprofitability

Lengnick-Hall et al., (2000) suggested that customer participation is the involving ofcustomer as an active participant in the work of the organization This view has two mainbenefits First, the integrated clients in customer engagement can help organizations reduce costs(e.g after customers use banking services, they discovered the irrational in the processtransactions, they raised these issues with bank managers, to collect opinions and adjustmentprocesses transactions of banks will help to reduce transaction errors and transaction time, thisobviously helps banks to cut staff costs, reduce transaction time, increased customersatisfaction) Besides that, customer participation can help reduce the price of products andservices (e.g cost and price are both normal covariates, the customer participation help bankscut costs as well means helping organizations improve profitability, and the considering servicediscounts will be considered in order to increase competitiveness, attract customers of the bank)

In line with this view, Mills et al., (1983) also said that customer participation lead to increaseproductivity for organizations Second, customer participation facilitates the bank to

customize its offerings to customers’ desires The private interaction between the consumer andthe bank consequently represents a vital part of a service delivery method within which theclient has direct input into the assembly of the ultimate service High-contact services, likemoney, legal, and medical services, are outlined by a high degree of coupling (i.e., continuous,immediate impact of contact), interdependence (i.e., service quality is conditional inputs fromevery party), and knowledge richness (i.e the worth of knowledge passed between parties)

(Kellogg and Chase, 1995) The advantages of client participation do not seem to be restrictedentirely to high-involvement services though; even in low- and moderate-involvement services,clients could notice customer participation enticing because they fancy magnified perceivedmanagement over the service delivery method (Bateson, 1985) and extra opportunities to createselections that offers them higher levels of customization (Schneider and Bowen, 1995)

Bendapudi& Leone (2003) and Ravald&Gronroos (1996) said that customer participationrepresent for things which the company could do with clients to co-create value Customers cansearch and see the benefits of it to make their impact on different parts of the worked system tobuild shopping and consumer experience of his own personal (Prahalad&Ramaswamy, 2004).Besides, customer participation not only helps the company consume products or services butalso help companies change and improve products/services quality Therefore, customers areincreasingly encouraged to actively cooperate with the company to create better value on thebasis of mutual benefit

According to Krajewski at al., (2010), the involvement of the customer is not always thebest option, but is accompanied by the disadvantages Allowing customers play an active role inthe service process may create trouble and makes the process less efficient The management oftime and volume of customer needs can be difficult and challenging Measuring service qualityalso becomes more difficult Costs for businesses could rise due to the layout of skilledpersonnel to better interact with customers However, the benefits of active participation in theprocess can increase customer value for themselves Some clients seek active participation andcontrol services process, especially if this participation saving both time and cost More client

involvement will mean higher quality, quicker delivery, bigger flexibility, and even lower price.Self-service is that the selection for several retailers, like supermarkets, gas stations and bankservices.

Encouraging customer participation will make the most customer competenciesthroughout the course of their innovation activities (Blazevic&Lievens, 2008) To inspirecustomers’ participation, enterprises may set-up and on-going relationship and work closely withcustomers to confirm meet customer needs During this sense, customer is thought as activepartners within the creation of innovative offerings Consequently, to amass superior servicequality for the client, corporations ought to be ready to develop new strategies (e.g socialcontrol and promoting innovations) to encourage customers to participate, in addition asstrategies to success monitor and manage the process along the way

2.3 Customer Loyalty

Customer loyalty refers to the trend to hold favorable attitudes to the services / productsand undertaking to acquisition and / or use of services / products (Bowen and Chen, 2001).Marketing researchers agree that in most of situation customer loyalty for a brand is moreimportant than customer satisfaction (Cooil et al, 2007; Pleshco and Baqer, 2008) In the otherhand, loyalty can make customers easy to forgive in a disappointing situation associated with acertain brand or to a store (Shankar et al., 2003)

In the opinion of Bendapudi & Berry (1997), even though the research literature ofcustomer loyalty is done with the different aspects, the two most popular views are the behaviorand attitudes In particular, the first approach of loyal is to meaning as repeat purchase behavior.However, there is also the idea that customers can eliminate repeat purchase behavior out oftheir habit because a number of barriers prevent them from changing suppliers or because of alack of attractive options or services instead In this respect, customers can maintain loyalty asobligations, developed a type of faked loyalty without any good feelings

with their suppliers The second approach is the considering of loyalty as a psychological state(emotional and or cognitive) that customers can achieve as a result of supplier’s relationalstrategies Customers may voluntarily maintain relationships based on the benefits they receive

in this relationship (Caruana, 2004) This approach supports the view that the foundation ofloyalty is actually a positive attitude to the supplier and ensures repeat purchasing behavior inthe future With the benefits that customers receive loyalty of customers to suppliers is derivedfrom themselves to maintain this relationship (Oliver, 1999)

In order to generalize the two approaches as Gremler at al., (2001) suggested, this studyconsiders loyalty to a particular supplier as a multidimensional concept In which, loyalty refers

to the extent to which customers expressed their intention to purchase again in the future(intention future behavioural), showing a positive attitude to suppliers (affective loyalty) and seevendors as an unique option for the needs arising in the future (cognitive loyalty) Derived fromtwo perspectives on loyalty, managers were given two main strategies to enhance customerloyalty The first strategy is to satisfy customers and provide them with the products and serviceswhich make they feel satisfied so that their desire to maintain this satisfaction The secondstrategy is to establish barriers to prevent conversion process of customer to a new provider(Balabanis, Reynolds, and Simintiras, 2006) If the customer is aware that the provision ofproducts and services on the market are homogeneous, there is little difference, they will not becombined with any benefits to changing providers The result is the tendency to abandon therelationship will be reduce and increase the level of loyalty (Balabanis et al.,2006)

2.4 Research model and hypotheses development

Building upon prior research, Figure 2.2 integrates the focal constructs of the thesis withhypothesized relationships

2.4.1 Bank choice behavior and customer loyalty

Nelson (1970, 1974) and Darby and Karni (1973) extended the concept of search,experience and belief in the economic information theory of Stigler (1961) by questioning howthe various types of attributes interact with the customer search and experience Accordingly,there is a distinction between search and experience attributes Search attributes is somethingthat customers can confirm before purchase via directly check or based on the source ofavailable information While experience attributes are things that customer can only get afterusing the product (Ford et al., 1990) On the other hand, according to Darby and Karni (1973),credence attributes are things that very difficult for customer to verify even after using Aproduct or service can have one, two or three types of attributes For example, in the case ofpersonal loan services: search attribute is interest rates, experience attribute is a professional inservice, and credence attribute will be integrity of staff Three types of properties are very useful

in checking customer disbelief (Ford et al., 1990, Smith, 1990), marketing information contents

(Norton and Norton, 1988), trust evaluation and confidence (Wright and Lynch, 1995), and theimportant role of price in making choice services (Ostrom and Iacobucci, 1995)

Customer loyalty usually is created after customers feel satisfied with the services thatbanks provide Therefore, customer loyalty will be not established before they experience theproduct In the case of search attributes (such as interest rate, overdraft privileges), consumersare able to inspect the services and draw inferences about the attribute before actuallypurchasing the service, and consumers are least skeptical of search claims (Ford et al., 1990)

Hoch and Ha (1986) showed that search attributes are relatively unambiguous in the theory ofinformation diagnosticity and ambiguity Therefore customers are most likely to review theircognitive framework that services quality of this bank is better or worse than that bank togeneric products while evaluating search attributes Therefore, this study hypothesizes that thechange of search attributes effect on customer loyalty

Hypothesis 1: There is a positive effect of search attributes on customer loyalty.

As discussed above, in opposition to aspects of search attributes, experience and credenceattributes are not easily confirmed before customers use a product or service that (Darby andKarni, 1973; Nelson, 1974) Also in the opinion of Kirmani and Wright (1989); Shapiro (1983);Wernerfelt (1988), customers must use a specific product / service to assess attributes and drawtheir own experience If customers want to evaluate the product before use, they will have torely on other features to draw conclusions about the quality of products/services Anotherproblem is that customers can not clearly evaluate credence attribute even after usingproducts/services So, in the absence of any other indication, customers tend to evaluatecredence attributes based on the customer perception Before using the service, customers oftenevaluate it based primarily on their basis perception of the service However, after the customerhas got the opportunity to try the service, they have extra information to associate with theirinference perception of its quality If a same services and differing only in name of bank, trialwould be expected to reduce the attitudinal loyalty for the service of bank have a low quality andincrease the attitudinal loyalty for the service of bank have higher quality Therefore, the studyhypothesizes that experience and credence attributes will have a positive effect on customerloyalty.

Hypothesis 2: There is a positive effect of experience attributes on customer loyalty Hypothesis 3: There is a positive effect of credence attributes on customer loyalty.

2.4.2 Bank choice behavior and customer participation

Experience represents a customer’s accrued knowledge about how a service shouldperform and general understanding of the average value of similar services in a differentprovider Experience likely increases client involvement of the customer for several reasons.First, as customers gain more experience in the type of product, they can better evaluate wherethey would make a contribution Moorthy et al (1997) similarly argue that as customers gainexperience, they are better able to evaluate the various attributes of different service offerings Inother words, customers who have a high level of experience are better equipped to make

valuable contributions to the production of a service and hence more likely to do so Second,inexperienced customers typically perceive higher decision-making risk (Heilman et al 2000).These customers are less likely to involve themselves because they fear producing a suboptimaloutcome Third, experienced customers likely have a greater need for control in the servicedelivery process (O’Connor and Siomkos, 1994) Therefore, the need to control or influence theoutcome of a service will increase expert customers’ likelihood of getting involved inproduction.

In the chain of bank choice behavior, customer joins all of the process from search,experience, to credence The interactive process creates the communication between client andbank staff Sharma and Patterson’s (1999) define communication as the formal and informalsharing of meaningful and timely information between a client and advisor in an empatheticmanner The content of communication, in the financial context of our study, focuses onencourage clients participate in the loan process, interacting with bank staff to search servicesinformation, response their understanding about the products and the responding of bank staff toclient for requesting information Communication flow is central to the establishment of strongrelationships; it builds trust by resolving clients’ queries and concerns and managing theirexpectations (Sharma and Patterson, 1999) Accordingly, author expects bank choice behaviorchain between the advisor and the client will increase the incidence of customer participationbecause of their greater willingness to share (potentially sensitive) information Accordingly, thestudy hypothesizes:

Hypothesis 4: There is a positive effect of search attributes on customer participation Hypothesis 5: There is a positive effect of experience attributes on customer

participation.

Hypothesis 6: There is a positive effect of credence attributes on customer participation.

2.4.3 Customer participation and customer loyalty

Customer participation is defined as constructive customer participation in the servicecreation and delivery process and clarify that it requires meaningful, cooperative contributions tothe service process In the delivery of financial services, such contributions might involve clientpreparation prior to meetings with advisors, client provision of input to decision makingprocesses, and timely and accurate responses to advisor requests for information

Customer participation likely yields several benefits for customers, including lowerprices, more opportunities to make purchasing, and larger discretion about the configuration ofthe ultimate product Moreover, clients likely experience shorter waiting times and enjoy agreater likelihood of customization The relationship between customer participation andperceived value, however, may not be unequivocally positive (Lengnick-Hall, 1996) Enterpriseshave less control over the training of customers to participate effectively in the productionprocess than they do with service employees In turn, customer involvement may raise theoverall level of uncertainty in the transaction and thereby affect outputs Furthermore, customersmay dislike the level of effort required to co produce a service and therefore abdicate their role

in the service process (Solomon, 1986) On balance, however, customers likely consider beinginvolved in a service creation process as a source of value, especially in the emerging era ofconnected, informed, empowered, active consumers (Prahalad and Ramaswamy, 2004).Dellande et al (2004) recently demonstrated that customer compliance-one facet of customerparticipation-relates significantly to customer satisfaction, and efforts by organizations toinvolve clients in customer participation lead to increased mutual understanding (Mohr andBitner 1991) Such mutual understanding results in positive emotional responses to the service(Price et al., 1995) Moreover, according to the view that satisfaction results from both cognitiveand affective assessments of service experiences (Westbrook, 1987), repeated positive emotionalexperiences ultimately contribute to the sort of customer satisfaction that underpins customers’decisions to be loyal to an organization (Lam et al., 2004) Therefore, the study incorporatemeasures of attitudinal and behavioral loyalty

(Dick and Basu, 1994; Wallace et al., 2004) Attitudinal loyalty refers to a measure of clients’intentions to stay with and level of commitment to the organization, whereas behavioral loyalty,

in the financial services context, is an objective measure of the amount of brokerage the clientpaid to the firm in the year following the administration of the questionnaire Taking thesemeasures of loyalty into consideration and on the idea of the preceding arguments, the studyhypothesizes:

Hypothesis 7: There is a positive effect of customer participation on customer loyalty.

Figure 2.2 - The proposed research model with hypotheses

CHAPTER 3 RESEARCH METHODS

This chapter discusses research methodology and methods for data collection and dataanalyses It presents the research design, development of survey questionnaires, qualitativestudy, and main survey This chapter also presents the data collection procedure and dataanalysis methods for testing the research hypotheses proposed in chapter 2

3.1 Research design

The research design of this study includes two phases The first phase, qualitativeresearch identifies the models, factors, suitable measurement variables for research in HCMC.Through the previous relevant researches, the questionnaire was built then running the pilot testfor checking the efficiency and the meaning of the questions The pilot test was purposed toexplore and define the relevant items and building a completed questionnaire The second phase,quantitative survey was the main approach of this study The goal is to examine the effect ofbank choice behavior and customer participation on customer loyalty in the personal loanservices of retail banking sector

Research process includes the steps as illustrated in Figure 3 (Steiger, 1990; Albright,2006; Schumacker at all, 2006)

Figure 3.1 - Research process 3.2 Development of questionnaire

3.2.1 Measurement scales

The measurement scales used in this study is multi-item five point Likert scales, whichwere well-established in the extant literature

3.2.1.1 Measure of search, experience and credence

Bank choice behavior consists of twenty-one items capturing three components (eightitems for search, seven items for experience and six items for credence) These items refer to theapplication of knowledge in bank choice behavior generating from the literature (Babakus

at al., 2004; Devlin, 2002) This construct will be measured on a five-point Likert scale withscale poles ranging from strong disagree to strong agree.

Table 3.1 - Measurement scales of bank choice behavior

variables

SA1 1 I always search interest rates on loans information of Interest rate

banks before I make decision of using services.

SA2 2 I usually ask banks what is the type of collateral the Collateral requirements

bank will accept to secure my loan?

SA3 3 I usually ask banks whether they willing offer me an Overdraft privileges

overdraft privileges account (unsecured).

SA4 4 I usually compare fee structure between banks when I Fees structure

attributes SA5 5 I always consider bank transaction hour whether it Banking hour

fixes with my times to do transaction at banks.

SA6 6 I usually make a research if banks have got new New bank services

services which have advantages for me.

SA7 7 I consider whether branch location near my house or Branch location

not.

SA8 8 I consider the bank that I use personal loan services is Image/reputation

a well-known or little-known EA1 1 Bank X’s staffs serve me with friendliness and friendliness and attentiveness

attentiveness.

EA2 2 Bank X provides services in a timely manner Timely manner of service

provision

Experience EA3 3 Bank X’s staffsare always willing to listen to me Willingness to listen

EA4 4 Bank X’s staffs present their professional in all expertise

Credence CRE1 1 I think that bank staff always has up to date knowledge Up to date knowledge of bank

CRE2 2 I feel that bank staff always show their integrity when Integrity of loan officer

I take transaction at bank.

CRE3 3 I see that bank staff has got knowledge of the market Knowledge about the market

when they converse with me.

CRE4 4 I believe that bank staff always keeping my private Information confidential

information in safely status.

CRE5 5 When I use services at my bank, bank staff usually Help customer up date new

introduces to me new technological advancements technological advancements CR6 6 I believe that my bank has enough competence in the Competence in the business of

3.2.1.2 Measure of customer participation

Customer participation consists of six items These items refer to the application ofknowledge in customer participation from the literature (Ngo & O'cass, 2013; Auh at al., 2007;Chan at al., 2010) This construct will be measured on a five-point Likert scale with scale polesranging from strong disagree to strong agree

Table 3.2 - Measurement scales of customer participation

variables

1 I spent a lot of time sharing information about my CP1 needs and opinions with the staff of bank X Sharing information

during the service process.

2 I put a lot of effort into expressing my personal CP2 needs to the staff of bank X during the service Showing needs

3 I always provide suggestions to the staff of Bank

X for improving the service outcome.

CP4 4 I have a high level of participation in the service Level of participation

using process of bank X.

5 I am very much involved in deciding how the CP5 services should be provided when dealing with Affect to providing services

bank X.

3.2.1.3 Measure of customer loyalty

Customer loyalty consists of four items These items refer to the application ofknowledge in customer loyalty generating from the literature (Auh at al., 2007) This constructwill be measured on a five-point Likert scale with scale poles ranging from strong disagree tostrong agree

Table 3.3 - Measurement scales of customer loyalty

variables

CL1 1 I will apply more loan amount at bank Xin the Willing to use more

future

Customer CL2 2 I will continue to use bank X for my needs in the Willing to continue using

CL3 3 The likelihood of me trying other services from Willing to try new services

bank Xis very good CL4 4 I try to use bank X for most of transaction needs Willing to use services

In summary, based on the previous research and after refinement, 30 variables are selectand group in fours indenpent variables and a dependent variable

3.2.2 Draft questionnaire

The questionnaire using a five-point Likert scale was employed to collect the data forfactors of the research model Items selects for the factors are mostly adapted from previousresearch in order to ensure content validity Items measuring search, experience, credence,customer participation and customer loyalty was take from previous research (see in 3.2.1.Measurement scales)

Structure of questionnaire consists of three main parts:

 Part 1: General information of customers who have used personal loan services This information helps select the target respondent to study

 Part 2: The main information includes statements (questions) are based on ascale of measurement was proposed for the research The items were measured

on the Likert 5-point scale from 1 to 5 (strongly disagree, disagree, neutral, agree, and strongly agree).

 Part 3: Questions on demographics characteristics are included at the end ofthe questionnaire

The survey questions were translate from English to Vietnamese by the researcher and asked to edit by others

The pilot test was conducted at Ho Chi Minh branch of Shinhan bank in Ho Chi Minhcity The 10 questionaires (translated to Vietnamese language) were sent to 03 banking expertsand 07 customers for answering All of them had much knowledge and many experience years inretail banking services After that, the forms have been return for the pilot test Some smallchanges on the questionnaire fits with understand of respondents and make clear for thequestions (in Vietnamsese language)

Finally, after adjusting the first questionnaire table, a small sample size of fifteenconvenient respondents was delivered the questionnaires to recognize whether any parts of itsunclear to understand or misunderstand However all of them understood meaning of

questionnaires quite well and knew the way to answer The last version of questionnaire wasestablished to survey in the massive areas After that, a main survey was conducted.

3.4 Sample method

The target population are the banking customers who are using personal loan of differentbanks in Ho Chi Minh City, the biggest city in Vietnam Due to limited time and cost, the targetrespondents are about 500 customers

Researcher will use the quota sampling method to collect research data It is a type of thenon-probability sampling It is mean that selected sample is not generalized to population.However, if we choose the right controlled characteristic (distinguishing attributes of studiedobjects), in theoretically, sample cannot represent for population but in practice it may representfor population The reason is when we divide the population according to controlledcharacteristics; the elements in the same group will usually have high homogeneity So, theelements involve in sample of selected groups although are selected in non-probability method,but because they have high homogeneity, so in some level, they can represent for population.Thus, the quota sampling method is a popular method which is usually used in scientific theoriestesting Even in applied research (collecting data to make business decisions), this method isalso commonly used, especially when the researcher do not know sampling frame (Nguyen DinhTho, 2012)

In this research, control characteristics including: location - Ho Chi Minh City; used to/are using personal loans services, gender, ages In which, the location characteristic wascontrolled by sending emails for person who has bank account in Ho Chi Minh city Used to/are using personal loans services characteristic was controlled in question form Yes/No Genderand ages characteristics were used to calculate exactly sample rate Researcher used the results

of population census 2009 in Ho Chi Minh City and the theoretical basis of quota samplingmethod to calculate the sample rate The results are as follows:

Table 3.4 - calculating sample with quota sampling method

on a large sample distribution theory (Raykov & Widaman 1995), how large of sample size isnot yet to be clearly defined Moreover, the sample size depends on the estimation method used(eg, ML, GLS or ADF).

Some researchers particularly considered with sample size can compute the desired Nrequired for a given model (e.g., some determined number of variables, constructs, and thereforedegrees of freedom) and desired level of power, or conversely, an estimate of power for a given

N (Kim, 2005; MacCallum, Browne, & Cai 2006) While others researchers found the addedbenefit that with three or more indicators per factor, a sample size of 100 respondents willusually be sufficient for convergent validity, and a sample size of 150 will usually be sufficientfor a convergent and proper solution” (Anderson & Gerbing, 1984)

In EFA analysis, the sample size is usually determined based on (1) the minimum sizeand (2) number of variables Hair et al (2006) suggested that in order to use EFA, the samplesize must be at least 50 and preferably 100 and the rate observations/ items should be 5:1 Itmeans that 01 measured variables need minimum of 5 observations (with 30 original observedvariables measuring customer loyalty, so to conduct EFA, the sample size of the study is at least30x5 = 150) If the method used to estimate is ML (Maximum Likelihood), the minimum samplesize from 100 to 150 (Hair et al., 1998) Thus, the sample sizes of the study should be at least

150 Using sample size n = 466 in this research is satisfy above conditions

3.5 Questionnaire administration

The respondents will submit questions through the online survey and direct survey Theresearcher sent 100,000 individual electric mails to whom have an account at Vietcombank/Shinhanbank and are living in Ho Chi Minh City and direct survey 50 persons at Vietcombank

Ho Chi Minh Branch After eleven day from 10/08/2014 to 21/08/2014, there are 550 customersanswered through online survey and 31 customers answered through direct survey However, tofix with the rate of gender and age from quota sampling method, the researcher only selected

466 respondents to conduct data analysis

3.6 Data analysis methods

All responded questionnaires were reviewed for completion, coded and input as raw data

in IBM SPSS Statistic The reliability and validity of measurement scales were evaluated byusing reliability analysis (Cronbach’s alpha test), exploratory factor analysis (EFA) andConfirmatory Factor Analysis (CFA) Then, analized method SEM is used to provide forinterpreting the results of its application from a managerial and statistical viewpoint The list ofcodified variables is shown in Appendix 1

CHAPTER 4 DATA ANALYSIS AND FINDINGS

The purpose of this chapter is to present finding which were collected from the actualquestionnaire survey Beside, the researcher proposed an official assessment of measures andalso carried out the analysis to give the accurate answers for the research questions, hypothesesproposing in the chapter 2

4.1 Descriptive analysis

The sample with 466 respondents included 48 men with year olds from 25 to 29, andothers as details in table below The percent is suitable with the selected sample rate of quotasampling method as describe in Section 3.4

Table 4.1 - Descriptive statistic of respondent’s characteristics

According to income variable, there were 70 persons have income equal or higher VND

20 million/month, counted 15%; 51 persons had income from 15 to lower VND 20million/month, counted 10.9%; 104 persons had income from 10 to lower VND 15million/month, counted 22.3%; 241 person had income from 5 to lower VND 10 million/month

There were 128 persons had postgraduate degrees, counted 27.5%; 324 persons hadbachelor’s degree, counted highest with 69.5%; 14 persons had finished high school level,counted 3.0% Most of respondent had high education level, it means that they easily understandthe questionnaires and answer it with high accuracy

Of 466 respondents, 136 persons in single status, counted 29.2%; 317 persons in marriedstatus, counted highest 68% and 13 persons in other status, counted 2.8%

Sample size = 466 observations

Name of variable Descriptions Frequency Percent Valid Percent

According to George and Malley (2003), Cronbach‟s alpha is used as criterion for

judging instruments or scales It only indicates if the items “hang together”; it does not

determine if they are measuring attribute Therefore, scales also should be judged on theircontent and construct validity George and Malley (2003) also provided the followingtechniques:

Table 4.3 - Cronbach’s alpha reliability coefficient

In testing of Cronbach's alpha, the observed variables have item-total correlation less than0.3 will be disqualified and criteria scale Cronbach's alpha coefficient to 0.6 or higher (Nunnally

& Burnstein, 1994) However, it should be noted that if the Cronbach's Alpha is too high (>0.95), it may appears redundant items in the scales

Table 8 depicts a summary of the beta scores of all the response ranking of the factorsthat affect to customer loyalty of person who use personal loan service The results indicated thatall scales satisfied the requirement for reliability All Cronbach‘s Alpha of the scales were higherthan 0.60 Among factors, “Experience attributes” factor has the highest ranking of Cronbach’sAlpha with valued 0.901 The lowest value of Cronbach‘s Alpha is 0.846 in factor “Customerloyalty”, the next valued is 0.866 in “Search attributes” and “Customer participation”, and 0.879

in Credence attributes In addition, the corrected item – total correlation in all scales satisfied thecriterion (> 50)

Table 4.4 - Cronbach’s Alpha coefficients for each measurement scale

Scale Mean if Variance if Item-Total Multiple Alpha if Item

Item Deleted Item Deleted Correlation Correlation Deleted

4.3 Exploratory Factor Analysis (EFA)

Norris and Lecavalier (2010, p.9) indicated that “EFA is based upon a testable model andcan be evaluated in terms of its fit to the hypothesized population model; fit indices can begenerated to help with model interpretation” The purpose of EFA test is to identify latentconstructs underlying a set of manifest variables”

According to Gerbing & Anderson (1988), Principal Axis Factoring Method with Promaxrotation (Oblique) will reflect data structure more accurately compare to Principal Componentsmethod with Varimax rotation (Orthogonal)

Hair at all (1998, 111) also suggested that Factor loading is the criteria of ensuringpractical significance Factor loading >0.3 are considered to reach a minimum, Factor loading

>0.4 are considered important, ≥0.5 were considered to have practical significance Hair at all(1998, 111) also advised that if selected standards Factor loading > 0.3, then sample size must be

at least 350, if sample size is about 100, the Factor loading should be > 0.55, if sample size isabout 50, the Factor loading should be >0.75

Moreover, factors with a total eigenvalue of 1 or greater will take into account; hence anyfactors with an eigenvalue of less than 1 are discounted (Lee & Hooley, 2005, p.376) Based onthese studies, any factors with eigenvalues greater than 1 would be retained And any factorloadings of 0.3 or higher on a factor are counted

KMO is a criteria used to consider the appropriateness of EFA, if 0.5≤KMO≤1 the factor analysis is appropriate Bartlett testing considers the correlation between the observed variables

to zero in population If this testing is statistically significant (Sig <0.05), the observed variables are correlated with each other in population So, researcher should be EFA with the following condistions:

- Using Principal Axis Factoring method with Promax rotation

- Factor Loading of every item must be > = 0.5

- In each item, the difference between | Factor Loading |'s largest and | Factor Loading | anyshould be > = 0.3 (Jabnoun & Al-Tamimi, 2003)

- Cumulative % of Extraction Sums of Squared Loadings > = 50% (Gerbing & Anderson,1988)

- KMO> = 0.5, Bartlett testing had statistically significant (Sig <0.05)

After testing Cronbach‘s Alpha coefficient, the measures were continued to be analyzed

by EFA analysis method (extraction method: Principal Axis Factoring and rotation method:Promax) After two time running EFA, the result indicated that there were five factors extractedand total variance extracted was 55.28% (greater than requirement of 50%) Thus, varianceextracted matched requirement However, there was 01 item (SA8) that not satisfied therequirements for factor loadings due to loadings < .50 So it was eliminated from themeasurement scales Factor Loading of all items were > = 0.50, the difference between largestFactor Loading and any Factor Loading was >= 0.30, KMO= 0.948>0.5, Bartlett testing hadstatistically significant (Sig = 0.000 <0.05)

Table 4.5 - The EFA analysis result

Extraction Method: Principal Axis Factoring.

Rotation Method: Promax with Kaiser Normalization.

a Rotation converged in 6 iterations.

Accordingly, EFA analysis modified the measurement scale with 05 factors, including:Search attributes (7 items); Experience attributes (7 items); Credence attributes (6 items);Customer participation (5 items), Customer loyalty (4 items) The scales had deleted-items fromEFA results, their Cronbach‘s Alpha were calculated again Consequently, the Cronbach‘s Alphatest and EFA results eliminated the unsatisfactory items in measurement scales The modifiedscales, which matched the requirements, were used to be tested by Confirmatory Factor Analysis(CFA) method

Table 4.6 - Cronbach’ Alpha results after excluding unsatisfactory items

Item-Total Statistics

Corrected Item- Squared Cronbach's Scale Mean if Item Scale Variance Total Multiple Alpha if Item Deleted if Item Deleted Correlation Correlation Deleted

4.4 Confirmatory Factor Analysis

To measure the degree of model fit to market information, people often use the square (CMIN); Chi-square adjusted for degrees of freedom (CMIN/df); Comparative Fit Index(CFI) Tucker & Lewis index (TLI); Root Mean Square Error Approximation (RMSEA) Themodel is considered appropriate for market data when tested with Chi-square P-value> 0.05.However, the weak point of Chi-square is dependent on sample size If a model receives valuesGFI, TLI, CFI ≥ 0.9 (Bentler & Bonett, 1980); CMIN/df ≤ 2, in some cases CMIN/df can be ≤ 3

Chi-(Carmines & McIver, 1981); RMSEA ≤ 0:08, RMSEA ≤ 0.05 are considered very good (Steiger,1990); so the model is considered to be consistent with market data or compatible with marketdata

Tho & Trang (2008) suggested that if the models get the value of TLI, CFI ≥ 0.9,CMIN/df ≤ 2, RMSEA ≤ 0.08, the model fit (compatible) with market data This rule is alsoused to assess the suitability of SEM model

In the first time run saturated model including 05 constructs (Search – Search attributes,Experience – Experience attributes, Credence – Credence attributes, CusPart – Customerparticipation, Cusloy - Customer loyalty), the results was pretty good with Chisquare/df = 2.153,GFI= 892, TLI= 936, CFI= 942, RMSEA= 05 Because Chi-square/df > 2, we would find theway to improve this index Based on the suggestions of Modification indices table, we woulddraw new variances between errors and run the model again

Table 4.7 - Modification indices

The result of second time running saturated model for 05 scale (standardized estimates)(Figure 4) Chisquare/df = 1.793, p = 000, GFI = 0.913, TLI = 956, CFI = 96, RMSEA =

.041 It demonstrated that these scales received an acceptable fit to the data (Figure 4) All thestandardized regression weights λ were substantial, greater than .50 (the minimum isstandardized λ SA3= 523 – Appendix 5) and significant (p < 05) The researcher will examinesome criteria before has official conclusions about the appropriate level of scale and researchmodel.

Figure 4.1 - CFA result in the second time (Standardized estimates)