Effects of Financial and Non-financial Performance Measures on Customers ’ Perceptions of Service Quality at Islamic Banks in UAE

This paper examines the influence of financial and non-financial performance measures on perceived Islamic banks’ service quality in the United Arab Emirates. External customers and internal customers of Islamic banks were considered in the study. External customers represent individuals dealing with the banks and their sample consisted of 230 customers. Internal bank customers represent banks’ employees with a sample size of 174 participants from five main Islamic banks in UAE. Data were collected mainly through a questionnaire and analyzed and tested through two proposed models using structural equation modeling techniques. The results indicate strong positive relationship between service quality and banks’ performance and that banks internal operations mediate the relationship for the external customers group.


Introduction
Performance measurement systems are necessary for every organization to evaluate its achievements (e.g., goals, satisfaction, resource utilization, etc.).With the increased level of globalization, strong competition, and technological changes, many companies have started to use a blend of financial and non-financial measures for their performance or what many writers would call balanced scorecards (e.g., Kaplan & Norton, 1996;Raphael & Man, 2013).As one of the tools used in the strategic process, balanced performance measures help the company assess its accomplishment of the strategic goals and objectives.Kaplan andNorton (1992, 1996) indicated that the financial measures of performance are insufficient to portray a full picture of a company's performance.Managers who rely only on financial performance measures usually end up with an incomplete view of what happened.On the other hand, the use of balanced performance measures that mix financial and non-financial measures can serve as a focal point of the organization's efforts that define and communicate its priorities to different groups of stakeholders (e.g., managers, employees, investors, customers, and the public).
As the next section on literature review reveals, a combination of financial and non-financial measures has become a popular framework in different fields such as economics, strategy, finance and accounting (Porter, 1992;Handy & MacDonald, 1994).This will provide a complete picture of banking performance from both quantitative (objective) and qualitative (subjective) perspectives.Performance measures strongly affect the behavior of managers, employees, customers, and investors (Handy & MacDonald, 1994;Kaplan & Norton, 1992).Although customer satisfaction is sometimes used as a proxy for service quality, the most well-known instrument to measure service quality is SERVQUAL (Parasuraman, Zeithaml, & Berry, 1985), which has five dimensions (Reliability, Responsiveness, Assurance, Empathy, and Tangibles).customers and internal customers (employees).The second step studied the relationship between customer perception of service quality and Islamic banks financial performance.
The remainder of the paper is organized into the following sections.The next section describes the literature review which is followed by the research method and model development.The fourth section reports on the results of testing for reliability and validity, the structural relationships, and the regression results.The paper ends with its conclusions, implications and limitations.

Literature Review
This section reviews relevant prior studies related to the use of financial measures, non-financial measures and service quality with special emphasis on the bank sector.

Financial Measures
The financial framework is the oldest paradigm for performance evaluation.Its roots are in the areas of accounting, financial management, and economics.Over the years, the accounting literature, for example, has recognized the importance of cost control, profitability, and liquidity.The literature documents the idea that reducing and controlling costs help increase profits.On the other hand, increasing revenues also helps boost profits when costs are controlled.
Financial and accounting reports employ various measures for profitability.They report profit as a gross amount (difference between revenues and cost of sale or services) or as a net amount (difference between revenues and total expenses).These amounts are absolute amounts and may not be appropriate for performance evaluation.Expressing these profit amounts as a percentage of some other base amounts seemed at a point in time to be more appropriate for performance evaluation.Consequently, performance evaluation systems have seen two new profitability measures: percentage of gross profit and rate of return on investment.
However, several authors (e.g., Kaplan & Norton, 1992;Ibrahim, 1999) indicated that although financial measures are important, they are not enough for a good performance evaluation system.The system should also incorporate non-financial measures of performance.One rational for this trend is provided by Kelly (2007).He indicated that firm value is created through different activities that promote critical success factors.These factors include innovation, quality, productivity, and customer satisfaction.Also Gu (2005) pointed out that these success factors ultimately improve future financial performance.Current summary financial measures that report financial results, such as operating income and return on investment, are unlikely to fully reflect the long-term consequences of these activities.Hence, many firms complement summary financial measures with nonfinancial measures that reflect key value-creating activities (Kaplan & Norton, 2001).However, Ittner and Larcker (2001) indicated that there is a lack of evidence on when and how nonfinancial measures improve managerial performance.

Non-Financial Measures
The use of non-financial measures to manage organizations appears to be positively associated with organizational performance on average.The association is often weak, however; there is considerable variation in the experience of individual organizations.Attempts to predict which types of organizations will benefit more from nonfinancial measures than others had mixed results (e.g., Hoque & James, 2000;Randall, 2003).
Another stream of studies focused on the role of strategic performance measurement systems such as balanced scorecards in assisting managers to develop competitive strategies.Chenhall (2005) indicated that a distinctive feature of such systems is that they are designed to present managers with financial and non-financial measures covering different attributes that translate strategy into a coherent set of performance measures.He identified three interrelated dimensions of integrative strategic performance systems.The first is the integration between strategy and operations.The second attribute is focusing on customer linkages.The third attribute is supplier orientation.Chow and Steve (2006) investigated the relative use of financial, nonfinancial and subjective performance measures derived from subjective judgment.They used a sample of 128 manufacturing firms.Their results indicated that companies with different manufacturing strategies use different mixes of the three types of measures.In addition, the different measures have different strengths and weaknesses (e.g., encouraging risk taking versus supporting decision making).Briggs, Claiboborne, and Cole (2006) pointed out that financial measures are generally lagging measures of performance while nonfinancial measures such as sustainability, learning and growth, and internal process improvements are leading measures of performance that offer insight about future performance.They estimated that investments affect gross operating income with a three-period lag but they affect nonfinancial measures without any time lag.Banker and Mashruwala (2007) assessed whether the reporting of nonfinancial measures such as employee satisfaction and customer satisfaction is likely to be useful to managers of a retail chain and under what competitive conditions such usefulness may take place.They found that employees and customers' satisfaction help managers in predicting future profitability.In addition, they found that customers and employees' satisfactions were correlated with the store closure decisions while customers' satisfaction only was correlated with performance evaluations of store managers.Dikolli and Sedatole (2007) presented taxonomy of empirical refinements of the leading indicators included in prior studies, all of which improve the information content of nonfinancial performance measures (NFPMs).These refinements incorporate: (1) alternative measurements of the NFPM, (2) lag differences in the leading indicator relation, (3) interactions between the NFPM and contextual variables affecting the leading indicator relation, nonlinearities in the leading indicator relation, and (5) variables that play a mediating role in the relation between the NFPM and future financial performance.Campbell (2008) examined the sensitivity of promotion and demotion decisions of lower-level managers to financial and nonfinancial measures of performance in a major fast-food retailer in US.He found that promotion and demotion decisions were sensitive to nonfinancial performance measures related to service quality and employee retention after controlling for financial performance.He also found evidence to suggest that lower-level managers behave in a consistent way with incentives created by assigning weights of nonfinancial performance measures.These findings provide some of the early empirical evidence on the use of weighted nonfinancial metrics in compensation contracts as a mechanism for generating improvements in performance.Brazel, Jones, and Zembelman (2009) investigate whether comparing financial data to nonfinancial measures (NFMs) can aid auditors and others in assessing fraud risk.They predict and found that fraud firms have greater differences in percent change in revenue growth and percent change in NFMs than their non -fraud competitors.These differences are positively associated with fraudulent financial reporting after controlling for variables that have been previously linked to fraud.

Service Quality
Much research effort and interest in service quality have been shown in the last thirty years or so (e.g., Oliver, 1980;Brown & Swartz, 1989;Carman, 1990;Bolton & Drew, 1991;Cronin & Taylor, 1992;Teas, 1993;Parasuraman, Zeithaml, & Berry, 1985, 1988, 1991, 1994;Foley, 2008, among others).This literature indicates that service quality is determined by technical aspects as well as the delivery of the service.Customers usually judge the technical quality based on their interaction with the service providers.However, quality of the service delivery is judged based on the customers' ability to access facilities, appearances, employees' behaviors, and the overall treatment presented to customers.This notion also extends to government services (Foley, 2008).
The service quality literature advocates that the main components of service quality are the five known dimensions (tangibles, reliability, responsiveness, assurance, and empathy).Tangibles refer to the physical appearance of the bank (its facilities, staff, and equipment).Reliability refers to the ability of the employees to perform the promised service dependably and accurately.Responsiveness refers to the willingness of the employees to help customers and provide prompt service.Assurance refers to the knowledge and courtesy of the employees and their ability to convey trust and confidence.Empathy refers to the level of care and individualized attention the employees provide to customers.Lewis, Orledge and Mitchel (1994) examined attitudes towards bank services among 147 university students.They measured students' expectations and perceptions of different bank features and services (e.g., loans, overdrafts, etc.).They used mean scores of students' expectations and perceptions to identify service quality gaps.They also used regression analysis to assess whether quality scores possess the ability to predict overall satisfaction.The results indicated higher students' expectations than their perceptions.Naser and Al-Khatib (1999) assessed the degree of customers' awareness of and satisfaction with the products and services offered by an Islamic bank in Jordan.They used a questionnaire to collect the data from 300 customers who had transactions at the bank's branches in 1998.The results indicated that a significant number of the participants were aware of the Islamic Bank's products and services.However, a small number of participants indicated that they were taking advantage of the bank's products and services.

Testing for Reliability and Validity
Reliability is the degree to which measures yield consistent results (Peter, 1979).Validity, on the other hand, measures the accuracy of an instrument (Stewart, 2009).We used cronbach's alpha as the measure of reliability.
For construct validity, we used different trials of confirmatory factor analysis (CFA).For validity testing; and based on the standardized loading and the error variance associated with each observed variable; we used three indicators: the average variance extracted (AVE) for convergent validity; the composite reliability (CR) to determine the internal consistency of a set of measures and discriminant validity (DV) which reflects the degree to which a dimension in a theoretical system differs from other dimensions in the same study (Churchil, 1979).
Table 2 shows all indicators of the non-financial model.The results in Table 2 showed that the CR greater than 0.7 the recommended threshold requirements for all dimensions; however, AVE values are acceptable if they exceeds 0.5 (Graver & Mentzer, 1999) therefore; two dimensions (Complaints and Empathy) will not considered in the external customers model.In addition, one dimension (responsiveness) will not be considered in the internal customer model.Also, it is worth to compare between the DV and the correlation values given in Table 4 and Table 5, it is very clear that DV values are larger than the correlation values within each dimension which indicate an evidence of discriminant validity (Fornell & Larcker, 1981).Moreover, the overall Cronbach's Alpha for both groups (External and Internal) are high and equal to .940 and 0.896; respectively; indicating a very good reliability of scale measurement.Furthermore, the items are regrouped into the optimal dimensions showing that customer satisfaction (Internal or External) with service in Islamic Banks in UAE is a multidimensional construct that allows explaining of 59.6 % of variance (External satisfaction) and 66.9% of variance (Internal satisfaction).
Pairwise correlation was also used to test the convergent validity (Anderson et al., 1987).If the lowest correlation of a particular dimension is significant at p < 0.01, convergent validity exists (Koufteros, Vonderembse, & Doll, 1998).In Table 3 and Table 4, the results show that the correlations values for each dimension are greater than 0.60 and 0.39 for external and internal models, respectively, and are significant at the 0.01 level.These results indicate initial evidence of good convergent validity.

Testing for the Non-Financial Structural Relationships
The research proposed models were tested using a linear structural equation modeling (SEM) with latent variables; which is well suited to highly complex predictive models (Jöreskog, 1973).SEM has several strengths that made it appropriate for this study, including its ability to handle both reflective and formative constructs.SEM analyses were performed using a covariance matrix as input to the Analysis of Moment Structure software package (Arbuckle & Wothke, 2003), using maximum likelihood estimation.The missing data were replaced by using the expectation maximization (EM) approach prior to analysis.For the evaluating of the model, residual means squared error (RMSEA), standardized root mean square residual (RMR), normed fit index (NFI), incremental fit index (IFI), comparative fit index (CFI) and goodness of fit index (GFI) values were taken into consideration.Noting that; a fit index value more than 0.90 and means squared error less than 0.08 would indicate a close fit of the model.The chi-square statistic for both models is significant with p value less than 0.05 and the ratio /df (see Table 5) drops well below the maximum value of 3 recommended by Kline (1998), indicating a good model fit.Moreover, it is worth to say that the fit indices given in Table 4 provide strong support for the external customer's model and most of these indices met the criterion but not the NFI (0.825) for the internal customer's model.
The estimated standardized regression coefficients are given in Table 5.These results are the answers of the first hypothesis, which indicate there is a positive relationship between customer satisfaction and all other dimensions for both models.Based on the standardized coefficients values it could be noted that the most important factor that effects on the external customer's satisfactions is the reliability (β = 0.982) then responsiveness (β = 0.946); tangibility (β = 0.909) and the least important factor is the assurance (β = 0.868).However, we observed different inference about the internal satisfaction preferences, where the complaint (β = 0.967) is the most factor effects on the internal customer's satisfaction, then empathy (β = 0.864), tangibility (β = 0.764), assurance (β = 0.727) and the least important is the reliability (β = 0.656).We also used the independent t test to compare between the customers (external and internal) satisfaction level (Table 6).The results indicated that there is a statistical difference between both groups, where the internal customers (Employees) are satisfied more than the external customers; in the bank service quality.Similar results also obtained in all non-financial dimension comparisons.
Tables 7 and 8 show the results of hypotheses 3, 4 and 5.It could be noted that the significant financial measures model differs between the internal and the external customers in terms of the service quality impact.While this dimension has a significant values and it has a positive impact on performances and on internal operations for the external customers, it has no significant impact in the case of internal customer models.In addition, there are similarities in terms of significant relations within all other measures, and it worth to say the standardized effects in the external customers are leading the same values in the internal customers model.The significant structural models for both groups are given in Figure 5. www.ccse tested the SERVQUAL and SERVPERF scales and the psychometric properties of the construct service quality in the Korean banking context.They used a 22-item questionnaire measuring

Table 2 .
Reliability and validity statistics for internal and external models

Table 3 .
Pairwise correlations and violations of discriminant validity for external customer's model

Table 4 .
Pairwise correlations and violations of discriminant validity for internal customer's modelNote.** Coefficient is significant at the 0.01 level.

Table 5 .
Regression coefficients of overall service quality

Table 6 .
Comparisons between internal and external satisfaction levels