THE EFFECT OF SERVICE PROCESS TYPE, BUSINESS STRATEGY AND ORGANIZATIONAL LIFE CYCLE STAGE ON THE DEGREE OF MCS BUREAUCRACY IN SERVICE ORGANIZATIONS

THE EFFECT OF SERVICE PROCESS TYPE, BUSINESS STRATEGY AND ORGANIZATIONAL LIFE CYCLE STAGE ON THE DEGREE OF MCS BUREAUCRACY IN SERVICE ORGANIZATIONS Sofiah Md. Auzair* Department of Accounting Faculty of Business Management Universiti Kebangsaan Malaysia Kim Langfield-Smith Department of Accounting and Finance Monash University Australia * Corresponding Author Department of Accounting Faculty of Business Management Universiti Kebangsaan Malaysia 43600 Bangi, Selangor Malaysia sofiah@pkrisc.cc.ukm.my 1

THE EFFECT OF SERVICE PROCESS TYPE, BUSINESS STRATEGY AND ORGANIZATIONAL LIFE CYCLE STAGE ON THE DEGREE OF MCS BUREAUCRACY IN SERVICE ORGANIZATIONS ABSTRACT This study uses the survey method to investigate the influence of several contingent variables on the design of MCS in service organizations. MCS was conceptualized in terms of five dimensions: action/results controls, formal/informal controls, tight/loose controls, restricted /flexible controls, and impersonal/interpersonal controls to form a composite measure of the degree of MCS bureaucracy.. The framework used in this study recognizes that the service process type, business strategy, and stage in the organizational life cycle influence the choice of MCS design within an organization. Questionnaires were administered to financial controllers of service organizations operating in Australia. Using t tests and multiple regression analysis, the results indicated that; (1) mass service, mature and cost leader firms place a greater emphasis on more bureaucratic forms of MCS, compared to professional service, growth and differentiator firms, and (2) service process type, organizational life cycle stage, and business strategies have a significant influence on the design of a firms MCS. INTRODUCTION Despite the dominance of the service sector in most economies, there has been limited research to investigate systematically the role of management control systems (MCS) in service organizations compared to the manufacturing sector (Shield, 1997; Drury, 1998; Sharma, 2002; Chenhall, 2003). It is not entirely clear why MCS in service organizations has received such limited attention. Addressing this concern, the literature (e.g. Brignall et al., 1991; Modell, 1998) frequently cites the distinguishing characteristics of pure services from tangible goods produced in manufacturing firms as posing important implications for control in service organizations. These characteristics are said to hinder the direct applicability of management accounting systems and MCS established in the manufacturing sector, to the service sector (Dearden, 1978). The unique characteristics of pure services include, (1) intangibility of services (2) inseparability of production from consumption, where customers are also involved in the production of most services, (3) perishability of services, where services not consumed will be lost forever, and (4) heterogeneity in service products, where services provided by the same person may differ between customers or differ at different times (Fitzgerald, Johnston, Brignall, Silvestro and Voss, 1991; Hope and Muhlemann, 1997; McColl, Callaghan and Palmer, 1998). From these characteristics, an important issue that emerges for the design of 2

MCS is the significant human involvement in the production process, which poses more uncertainty for the control of service production. Several case studies have demonstrated the process and design of MCS in service organizations, but it has been suggested that the traditional role of MCS expounded in manufacturing organizations requires a re-orientation to be effectively implemented within service organisations (e.g., Abernethy and Stoelwinder, 1991, 1995). These studies imply that service organizations are unique and require special attention. Nevertheless, Sharma (2002) noted that much of the research undertaken in the service sector has focused on non-profit organizations. Thus, the need to re-orientate MCS suggested by these studies may stem from differences in the nature of operations and the organizational context between manufacturing and non-profit service firms, rather than from differences in the characteristics of service versus manufacturing organizations. The shared goal of profitability would make profitoriented service organizations like manufacturing firms, more likely to implement and utilize MCS (Sharma, 2002). This suggests that it is possible to approach studies of MCS design in service organizations based on this common ground, rather than merely focusing on their uniqueness. Prior studies of MCS in service organizations usually focus on organizations in a single industry (e.g., the health care industry or communications industry) or on a single organization (e.g., Marginson, 1999). As service industries range across retail, health care, education, hospitality and professional services, limited generalizations can be drawn from such a narrow focus. Thus, it is hardly surprising that MCS research in service organizations is regarded as only in its infancy (Modell, 1996, 1998). In addition, current knowledge of how management control operates in various service contexts is also relatively limited (Modell, 1998). The purpose of this paper is to add to the limited body of knowledge in this area. Specifically, This paper will identify the influence of specific organisational factors that influence the design of MCS in service organizations across several different service industries. The remainder of this paper is organised as follows. First, the conceptual framework that forms the focus for this study is presented, followed by formulation of a series of testable hypotheses. In the following section, the research method is outlined detailing the sample and measurement of variables. The results of the study are then provided. The findings of the study are then reviewed while major themes are discussed. In the final section, the limitations of the study are presented and areas for future research are suggested. 3

CONCEPTUAL FRAMEWORK The service process type, business strategy and organizational life cycle stage have been identified in prior studies as having an effect on the design of MCS design in service organizations (Brignall, 1997; Brignall et al., 1991; Kumar and Subramaniam, 1997; Moores and Yuen, 2001). Nevertheless, these variables have never been studied in a single study across a variety of service firms. While there have been published studies utilizing the service process model, this has not been used in quantitative studies to examine the linkage between service process type and the design of MCS (e.g. Fitzgerald et al., 1991; Brignall et al., 1991; Brignall, 1997; Sharma, 2002). Business strategies have been studied extensively in the MCS literature, however, none of these studies focused on the relationships in several different service industries. With regard to organizational life cycle, the only empirical evidence relating this variable to MCS had been in clothing and footwear industry ranging from pure manufacturing to retail outlets (Moores and Yuen, 2001). In addition, no published study to date has considered all the variables simultaneously. The framework proposed in this study therefore, remains exploratory. Exactly which of these variables will significantly influence the MCS design in service organizations is subject to empirical investigation. Accordingly, in this study, a framework as presented in Figure 1 is proposed. The relationship stems from the recognition that as an organization strives to achieve effectiveness, it will seek to attain fit between the contingent variables and the MCS, thereby creating relationships between these variables and the MCS. PLACE FIGURE 1 HERE MCS There is limited consistency in the way that MCS have been characterised in the prior literature. Some researchers focus on single aspects of MCS. These include, for example, action versus results controls (Merchant, 1998; Ouchi, 1979), and formal versus informal controls (Modell, 1995; Amigoni, 1978). Concerns have been voiced on the limited number of contingency-based MCS studies examining broader elements of control, which subsequently leads to serious model under-specification (Chenhall, 2003). It has also been argued that the variation in the number and type of controls that have been researched makes it difficult to develop a coherent body of knowledge (Langfield-Smith, 1997). Amigoni (1978) suggested that some distinctive features of MCS should be identified when proposing a conceptual framework useful in designing and implementing MCS in any specific organization. 4

Several distinctive features of the MCS can be identified from prior studies (see Ouchi, 1979; Modell, 1995; Merchant, 1998; Whitley, 1999; Chenhall, 2003) to characterize MCS. They will be referred to as MCS dimensions as the term dimension is seen as most appropriate in explaining the broad scope covered by each attribute. Each dimension and their respective sources are presented in Table 1. Including these dimensions in a MCS framework is expected to provide a broad perspective of the system. In addition, it has been suggested that classifying controls along a bureaucratic continuum might be useful for addressing concerns of how MCS relate to broader control systems (see Chenhall, 2003). PLACE TABLE 1 HERE Organizations are expected to exercise control along a continuum for each of the MCS dimensions outlined in Table 1. Referring to definitions (Burns and Stalker, 1961; Perrow, 1970; Chenhall, 2003) and descriptions of bureaucratic controls used in prior studies (Abernethy and Stoelwinder, 1991, 1995; Ouchi, 1977, 1979; Whitley, 1999), it is suggested that on one end of the control continuum is the more bureaucratic - action, formal, tight, restricted, and impersonal controls. Accordingly, the other end of the continuum is the less bureaucratic - results, informal, loose, flexible, and interpersonal controls. Service Process Type Several prior studies have identified the service process type as a potentially important variable for future MCS studies (e.g, Brignall, 1997; Brignall, Fitzgerald, Johnston and Silvestro, 1996; Modell, 1998). While the service process type variable defines the characteristics of service organizations, in this study, it is also regarded as a variable providing a distinct contingency for the adoption of MCS. This suggests a shift from a focus on the development of typologies of service organizations, towards the use of the characteristics of these typologies to suggest relationships with MCS. The service process type variable is adopted from the service process model proposed by Silvestro et al. (1992). Silvestro et al. (1992), proposed a model that took into consideration the essential characteristic of service industries, which is the existence of the customer within the production system. Using data from case studies of 11 for-profit service organizations, each company was ranked based on the volume of customers processed by a typical unit per day, and classified against six dimensions drawn from the service operations literature, where the definitions of the dimensions were examined, developed and refined. Table 2 summarizes the six dimensions and their definitions. 5

PLACE TABLE 2 HERE The six dimensions have, over several decades, become widely recognized and used in the service operations literature. Silvestro et al. (1992) extended prior studies by combining all six dimensions, to propose three service archetypes. They are professional services, service shops and mass services. The definitions of the three service archetypes are as follows: 1. Professional services are defined as organizations with relatively few transactions, highly customised, process oriented, with relatively long contact time, with most value added in the front office, where considerable judgment is applied in meeting customer needs. Examples are management consultancy and corporate banking. 2. Mass services are organizations where there are many customer transactions, involving limited contact time and little customisation. The offering is predominantly productoriented with most value being added in the back office and little judgment applied by the front office staff. Examples are confectionery, tobacco and newspaper retailers and transport. 3. Service shop is a categorization which falls between professional and mass services with the levels of the classification dimensions falling between the other two extremes. Examples are hotel and rental services. By considering the various dimensions together in classifying services, this scheme is not only a service classification scheme but is also able to offer strategic insights. As common characteristics between firms in the service industries are identified, managers would be able to focus on managerial issues across similar service industries and analyse their position in the market place. Silvestro et al. (1992) further proposed that the three types of service processes give rise to different management concerns, and thus control and performance measurement will differ significantly between the three service types. The service process type is viewed as spanning on a continuum, where at one end of the continuum are mass service firms and at the other end are professional service firms. Thus, in this study, the contrast between the MCS of these two extremes will be addressed. Business strategy Business strategy is concerned with how businesses achieve competitive advantage (Slater and Olson, 2001). It has been suggested that the MCS should be tailored explicitly to support the strategy of the business to lead to superior performance (Langfield-Smith, 1997). 6

Porter (1980) proposed that regardless of industry context, organizations can choose from one of three generic strategies to compete at the business level. They are cost leadership, differentiation and focus strategies. Cost leadership requires aggressive construction of efficient-scale facilities, vigorous pursuit of cost reductions from experience, tight cost and overhead control, avoidance of marginal customer accounts, and cost minimization in areas like R&D, service, sales force, advertising and so on. Differentiation refers to creating products or services that are perceived industry wide as unique. Some of the differentiation approaches include quality of the product, wide availability of product offerings, product flexibility, technology and customer service. Focus aims used a cost leadership or differentiation approach to concentrate on a particular buyer, segment of the product line, or geographic market, thus servicing its narrow strategic target more effectively or efficiently than competitors who are competing more broadly. One consideration in selecting the appropriate strategic typology for this study is the need to be able to operationalize the concept in a manner that is easily understandable by managers participating in this research. 1 According to Miller and Dess (1993), Porter s typology provide a kind of shorthand that encapsulates complex, holistic gestalts in a simple language that is seen as most appropriate to achieve the aims of capturing broad view of MCS and unravelling the complex relationships expected to exist in service organizations, while maintaining simplicity. With regard to simplicity, Porter s typology has been commonly used to address the differentiation and cost leadership strategies, as convenient labels describing a complex gestalt of strategic attributes (Miller and Dess, 1993). According to Langfield-Smith (1997), competitive advantage is based on the cost leadership or differentiation strategy. In this study, the contrast between these two strategies and the implications for the design of MCS will be addressed. Organizational life-cycle stage The potential influence of the life cycle stage on MCS and organizational effectiveness, and the relatively scant attention received in the literature, provides an opportunity for the current research to include the life cycle stage as one of the contingent variables. A firm s stage of growth represents a contingency or driving force to which appropriate organizational 1 Porter s typology was considered an appropriate means of conceptualizing business strategy in this study. Frequently mentioned is the congruence between Porter s and Miles and Snow s (another dominant business strategy typology in MCS literature) typologies (Shortell and Zajac, 1990). For instance, Porter's overall cost leaders are similar to Miles and Snow's (1978) defenders and Porter's differentiators are comparable to Miles and Snow's prospectors (Kumar and Subramaniam, 1997). Porter s typology is also comparable to the strategic mission of build and harvest (Langfield-Smith, 1997). The similarity of Porter s typology to other strategic typologies, could facilitate comparability of research finding (Kumar and Subramaniam, 1997). 7

responses must be matched (Miller and Friesen, 1984). Brignall (1997) argued that service control systems might need to change throughout the life cycle to fit in with changes in the competitive environment, business mission and generic strategy. Failure to account for the stages of the life cycle in budgetary policies has been found to result in inefficiencies of resource allocation (Czysewski and Hull, 1991). A review of the literature suggests that the organizational life cycle variable is more researched in organizational studies, compared to MCS studies. Among those few MCS studies that do address life cycle, the approach has been more conceptual than empirical (see for example, Moores, 1990; Brignall, 1997). To the knowledge of the authors, the only published empirical work in the MCS area is Moores and Yuen (2001). They adopted Miller and Friesen s (1984) life cycle model to study management accounting systems (MAS). According to Miller and Friesen (1984), life cycle stages imply integral complementarities among the environment, strategy, structure, and decision-making methods, while each stage exhibits certain significant differences from all other stages along these four elements. Miller and Friesen suggested five stages of life cycle: birth, growth, maturity, revival and decline stages. In fact, a number of multistage models have been proposed in the organizational life cycle literature, ranging from three-stage to five or more stages models (Kazanjian, 1988). Reviewing much of this literature, Quinn and Cameron (1983) summarise nine life cycle models to develop a four-stage life cycle model. The stages and the corresponding characteristics are summarized in Table 3. Although the nine models are based on different organizational phenomena (e.g., structure, individual mentalities, functional problems), they all suggest similar life cycle stages and there is some consensus about the characteristics of certain stages (Quinn and Cameron, 1983). PLACE TABLE 3 HERE As only one of the nine models included the decline stage, the model integration was not complete (Quinn and Cameron, 1983). However, it was argued that the stages from birth to maturity might be the only predictable ones as in mature organizations (after the elaboration of structure stage) life cycle models break down, and change occurs metamorphically and unpredictably. Nevertheless, to a certain extent, the summary by Quinn and Cameron (1983) provides comprehensive insights into the characteristics of firms in different organizational life cycle stages. 8

HYPOTHESES FORMULATION Not all of the MCS dimensions used in this study have been explicitly addressed in prior studies that investigate the relationships between selected contingent variables and MCS design. However, relationships with the aggregated MCS (all MCS dimensions) are expected to exist, and will be based on the similar more and less bureaucratic character shared by these dimensions. Logically, if all MCS dimensions are expected to span across the bureaucratic continuum, the relationships expected of the contingent variables might also be influenced by this criteria. Therefore, some logical link may be formed between the contingent variables and the MCS in situations where no prior studies have explicitly addressed the relationships. Earlier it was mentioned that although there exists some differences between the pure service provider and manufacturing firms, profit-seeking service organizations are similar to manufacturing organizations in emphasizing the goal of profitability and meeting performance targets. A growing body of literature demonstrates that MCS are equally important in service organizations and that MCS of service organizations must be considered in their wider context, which includes accounting for contingent variables external and internal to the service organizations (see Sharma, 2002). Therefore, while adjusting for the differences in MCS design expected in service organizations, some empirical evidence found in the manufacturing setting may also be considered in forming the testable hypotheses. Service process type The process types in this study are viewed as spanning a continuous spectrum: one end is the mass service firms and the other, are the professional service firms. Based on the dimensions used to characterized these process types, mass service firms are more likely to face less uncertainty, while professional service firms are more likely to face high uncertainty. For example, for the equipment/people dimension outlined in Table 2, the involvement of more human aspects in the production processes of people-based organizations such as a consultancy firm, produces a higher degree of uncertainty compared to equipment-based organizations such as transportation. Prior studies have found that when the task is uncertain, results control are more likely (Eisenhardt, 1985). According to Eisenhardt (1985), the task, which refers to selling task in retail services, is found to be a principal determinant of whether commissions (a form of results control) or salaries (a form of action control) are used. Notably, the relationship expected between action and results control with service process type discussed above was inferred, based on the evidence found between the latter and task 9

uncertainty 2, a variable often studied in the MCS literature. The inference is considered reasonable based on the uncertainty element reflected in the service process type dimensions. Modell (1998) warned that considerable caution would be required when extending studies using the task uncertainty dimension to the service context as the instrument used to measure task uncertainty lacks direct reference to customer activities, an important element that presents the uncertainty in service organizations. Modell did, however, acknowledge the partial similarities which could offer some scope for meaningful comparisons. The direct link between MCS and the service process type can, however, be found in case studies undertaken in several service organizations by Fitzgerald et al. (1991). This study includes explicitly a comparison of performance measures implemented in organizations of different service process types. The measures of quality in professional service firms, as found in Fitzgerald et al. (1991), focused on customer satisfaction, measured at the individual customer level. In mass service firms, on the other hand, the relationship was between the customer and the organization as a whole. With less direct responsibility for quality and less discretion at the front line, many more structured quality control systems are required for mass service firms, compared to professional service firms. Consistently, their findings suggest that service quality measurements in mass service firms are structured and formal, while unstructured and informal measurements are found in professional service firms. In terms of the scope for the provision of service flexibility, Fitzgerald et al. (1991) found that professional service firms appeared to have higher degree of flexibility than the mass service firms. This indicates that an MCS that allows quick responses is more critical in professional service firms, if they are to operate efficiently. Professional services need flexible job scheduling systems to support the negotiation of delivery dates with customers, and the high degree of customisation offers greater specification flexibility. Highly skilled contact staff, multi-skilling and cross training are commonly used to maximize flexibility in assigning staff to jobs. In mass service firms on the other hand, services are more standardized. Thus, specification flexibility tends to be a matter of providing a range of options from which customers can choose (Fitzgerald et al., 1991). 2 Tasks are uncertain when they are high in variety and low in programmability (Perrow, 1970). Similarly, as professional service firms are involved in highly customized service offerings and long contact hours, they will involve a high degree of variability in the task performed and the task is less programmable. Professional service firms have a high degree of task uncertainty. In contrast, mass service firms are involved in more standardized service offerings and require less direct contact with customers. Consequently, they face less variety in the tasks performed and the tasks become more programmable. Mass service firms have low task uncertainty. 10

A study undertaken in public accounting firms suggests an association between the professional service process type and the impersonal and interpersonal control style. In considering the leadership style, Otley and Pierce (1995) suggested that the working environment uncertainties made a highly considerate style more helpful than a structured style in public accounting firms. 3 The auditor is exposed to boundary-spanning activities to a greater degree than many other professionals, giving rise to additional layers of uncertainty. A structured leadership style may offer little or no role clarification, while a considerate style promotes a supportive work environment in which the uncertainties can be addressed effectively. As the considerate leadership style closely resembles the interpersonal emphasis and a structured style could be perceived as more impersonal style of the MCS design, it could be inferred that professional services (of which public accounting firms are an example) would be more likely to emphasize an interpersonal form of control than mass service firms that relatively face less task uncertainty. Although no specific empirical findings address the tight and loose control dimension, it has been suggested that the form of control varies with the nature of the concern s task (Burns and Stalker, 1961; Perrow, 1979; Woodward, 1965). When tasks are easy to define, tightening the controls may provide a means for ensuring that people are acting as the organization wishes. On the other hand, when tasks are highly uncertain, these controls may prohibit creativity and could result in dysfunctional behaviour. Thus, some controls have to be loose to accommodate uncertain work demands if the firm is to operate efficiently. In summary, the findings of these studies suggest that professional service firms are more likely to emphasize results, informal, loose, flexible, and interpersonal controls (which represents a less bureaucratic form of MCS). Accordingly, as mass service firms are positioned at the other end of the continuum, a more bureaucratic control is expected to be more suitable. This discussion therefore leads to the following hypothesis: H 1 : Mass service firms will place greater emphasis on a more bureaucratic form of MCS than professional service firms. 3 A structural style of leadership is associated with perceptions of tight control and inflexible use of the time budget. A more considerate style of leadership tends to be associated with mutual trust and respect, helpfulness and friendliness (Otley and Pierce, 1995). 11

A summary of the expected relationships between service process type and specific dimensions of MCS that were derived from prior empirical research and from theoretical reasoning is contained in Table 4. PLACE TABLE 4 HERE Cost leadership and differentiation strategies With regard to the organizations business strategies, a review of the prior literature indicates numerous empirical studies have used a contingency approach to examine the relationship between specific aspects of MCS and business strategy (see Langfield-Smith, 1997, for a review). Kumar and Subramaniam (1997) found that cost leaders place a greater emphasis on formulating effective personnel policies, which is comparable to action controls, while the differentiator places highest emphasis on rewarding creativity which is comparable to results control. Several research findings suggest that differentiation strategies are associated with informal, loose and flexible controls (Govindarajan, 1988; Van der Stede, 2000; Sim and Teoh, 1997; Simons, 1987). These control features allow the differentiator to be innovative in creating products that are perceived as unique by customers. Cost leaders, on the other hand, have been found to be positively associated with formal and restricted controls (Miller, 1988). However, contradictory findings were suggested by Simons (1987) when he reported that effective prospectors use their financial control systems more intensively than defenders, with tighter budget goals and more frequent reporting, and placing more emphasis on forecasts and monitoring output than on cost controls. No significant relationship between cost control and performance was found for defenders. The contradicting and fragmentary results of this empirical study have been acknowledged in several literature reviews (see for example, Dent, 1990; Langfield-Smith, 1997; Sim and Teoh, 1997), yet the explanation remains speculative. However, these reviews suggest that the results of most studies do not provide support for Simons findings. There is no published empirical evidence that directly links business strategy to the impersonal and interpersonal control dimension. Therefore, the nature of cost leadership and differentiation will be considered to establish the link between business strategy and the MCS design. Cost leaders actively pursue cost reduction, tight cost and overhead control and cost minimization. This suggests that cost leaders are likely to adopt uniform rules to focus employees on tasks and prevent them from working outside strict boundaries. An impersonal 12

type of control is therefore more suitable for firms pursuing cost leadership strategy. On the other hand, as differentiators need to create services that are perceived industry wide as unique, they need to allow employees more space to be creative and flexible. An impersonal approach to control may result in dysfunctional behaviour when employees perceive it as limiting their ability, and in turn may affect their creativity. Therefore, an interpersonal approach to control is more likely to be consistent with a differentiation strategy. To sum up the discussion, a differentiation strategy is associated with an MCS characterised by results, informal, loose, flexible, and interpersonal controls, which are features of a less bureaucratic MCS. On the other hand, a cost leadership strategy is associated with action, formal, tight, restricted and impersonal controls, which are features of a more bureaucratic MCS. The preceding arguments suggest the following hypothesis: H 2 : Firms pursuing a cost leadership strategy will place greater emphasis on a more bureaucratic form of MCS than firms pursuing a differentiation strategy. Table 4 contains the summary of the relationships between business strategies and MCS dimensions derived from prior and theoretical arguments that formed the evidence for H 2. Organizational life cycle With regard to the organizational life cycle, Miller and Friesen (1984) tested empirically the features of strategy, structure, environment and decision-making style of the organizational life cycle stages by employing a longitudinal method of analysis. The results indicate complementarities among variables within each stage and inter-stage differences were evidenced. With regard to control systems, they found that firms in the maturity stage would place more emphasis on formal cost controls and budgets. Moores and Yuen (2001) utilized Miller and Friesen s organizational configuration concept to study MAS in the clothing and footwear industry. Contrary to most life cycle literature (see for example, Kimberly, Miles and Associates, 1980), Moores and Yuen (2001) expected that MAS formality would be highest at the growth stage, and relaxed at maturity. Their results supported their hypothesis. However, according to the details of the summary of the nine life cycle models by Quinn and Cameron, it is highly likely that as a firm moves from the growth stage to the maturity stage, the controls will become more formal. At the growth stage, firms are characterized by their innovative features that emphasis on quick response and thus, too much formal control would stifle flexibility and creativity. Accordingly, Burns and Stalker (1961) suggested that innovation is more suited to unstructured and organic organizations, where there is less reliance on formal controls. 13

As the organizational life cycle is a fairly recent variable in the MCS literature, it is not surprising that the stage in the life cycle has not been linked empirically to most of the MCS dimensions. However, Miller and Friesen (1984) found that generally, firms at the growth stage are associated with a bold, innovative and organic orientation. Mature firms, which are stable and consist of normally larger organizations, demonstrate conservative practices, encroaching on bureaucracy. In summary, this implies that growth firms are more likely to adopt less bureaucratic controls compared to mature firms. Based on this reasoning, the following hypothesis is proposed: H 3 : Firms in the mature stage of their life cycle will place a greater emphasis on a more bureaucratic form of MCS than firms in the growth stage. Multiple contingencies The discussion above suggests a piecemeal approach that links the contingent variables to MCS in service organizations. These organizations, however, need to face various situations concurrently, and effectively the MCS design adopted is a result of considering multiple contingent situations simultaneously. Notably, researchers have been selective in studying the relationships between the contingent variables and the MCS, probably to gain tractable findings. However, Gresov (1989) argued that focusing on a single situation may fail to capture the design process and consequently lead to a lack of widely accepted theory of management control. In the first three hypotheses, individual contingent variables were expected to influence the MCS design. Logically, a variable that influences the MCS of an organization individually should also influence MCS in that organization in combination with other variables. The task is to establish the most appropriate link with MCS when all variables are considered simultaneously. Earlier discussions accompanying Hypotheses 1 to 3 demonstrated that each contingent variable argued for a a direct relationship with MCS, and it is expected that a simultaneous direct relationship between all contingent variables with MCS would also exist. The possible effects of these contingent variables on MCS design can be deduced from what is expected from the hypothesised relationships between individual contingent variables and MCS design. However, it is noted that in considering all variables in a single analysis, some variables may reduce in the significance of their relationships with MCS, in the presence of other variables. 14

The discussions on the individual contingent variables relationship with the bureaucratic form of MCS above indicate that a more bureaucratic MCS is more likely to be adopted by mass service firms that follow a cost leadership strategy and are at the mature life cycle stage. On the other hand, a less bureaucratic MCS is more likely to be adopted in professional service firms with a differentiation strategy and at the growth life cycle stage. Table 5 summarises the expected relationships between all contingent variables for firms adopting a more bureaucratic MCS and less bureaucratic MCS. PLACE TABLE 5 HERE Based on the foregoing discussion, the following hypotheses are proposed. H 4a : Mass service, cost leaders, and mature will be more likely to adopt a more bureaucratic form of MCS. H 4b : Professional service, differentiators, and growing organizations will be more likely to adopt a less bureaucratic form of MCS. RESEARCH METHOD Sample and survey procedure Survey questionnaires were administered to 1,000 financial controllers of service organizations operating in Australia. A four-page questionnaire was constructed as part of a larger study. The questionnaire was refined based on the feedback of a pre-test with 20 service company managers and 10 academic staff representing peer reviews. financial controllers were targeted as the respondents as they were expected to have the best knowledge of the various aspects of their organsitaions, but in particular knowledge of the MCS. A mailing list for the service organizations was obtained from a commercial mailing list company, which covered several different service industries. The industries in which these organizations were classified, matched the service industries classification provided by the Australian Bureau of Statistics, and the industries selected by Silvestro et al. (1992) to classify organizations by service process type. A total of 155 (15.5 per cent) responses were received of which 149 (14.9 per cent) were usable. A data screening procedure was undertaken to analyze missing responses and outliers. 15

This step further reduced the sample to 121 responses. Table 6 lists the final sample distribution according to service industries categories. PLACE TABLE 6 HERE The low response rate obtained from the survey has raised the potential for non-response bias. Often non-responders are different in crucial aspects to responders (de Vaus, 1985), therefore, to allow some degree of generalization a non-response bias test were undertaken. A comparison of the means of all variables measured on Likert scales were undertaken which included all items used to measure service process type, MCS, cost leadership and differentiation strategy. The life cycle stages were measured as a categorical variable, and thus, were not tested. The results indicate no significant differences (p 0.05) for responses for all the questions on service process type, MCS, cost leadership and differentiation strategy. Measurement of variables MCS The instrument used to measure the five dimensions of MCS consisted of a series of semantic differential scales. A brief explanation of the nature of the controls for each dimension was provided on the opposite ends of each scale, representing a forced-choice format. Wherever possible, questions were adopted from prior survey instruments and modified for use within the current study (Robbins and Barnwell,1989; Simons, 1987; Govindarajan and Fisher, 1990; Ouchi, 1977). Respondents were asked to indicate the extent to which the two anchored statements best described the type of MCS emphasized in their organizations by choosing a number on a 7-point scale. The lower end of each scale (score 1) indicates a greater emphasis on more bureaucratic MCS. The higher end of the scale (score 7) indicates a greater emphasis on less bureaucratic MCS. Although based on some theoretical foundation, no published study has tested whether these items when grouped together, measure the same underlying concept, especially in the service sector. To determine whether these items belong to the same scale, the items were factor analysed to test for unidimensionality (de Vaus, 1995; Hair et al., 1995). Principal components analysis (PCA) was selected as the extraction method. PCA is used when the objective is to summarize most of the original information (variance) in a minimum number of factors for prediction purposes (Hair et al., 1995). 16

The Bartlett test and the KMO measure of sampling adequacy indicate suitability for factor analysis. All items load well in excess of 0.32 (as recommended by Tabachnick and Fidell, 2001) and therefore were retained in the analysis. The reliability of the scale was.62, just above the lower limits of acceptability for exploratory research, generally considered to be around 0.50 to 0.60 (Nunnally, 1978). The variance explained by the scale was 40.39 per cent. PLACE TABLE 7 HERE Service Process Type Six items representing six dimensions based on the typology developed by Silvestro et al. (1992) were measured on a 7-point scale, by providing a series of brief explanations, which formed anchors for the semantic differential scale. The lower end of the scale (score 1) indicates a greater emphasis on mass service process type while the higher end of the scale (score 7) indicates a greater emphasis on professional process service type. There has been no published study that has developed a survey instrument to measure the service process type. Thus, the combination of these items needs to be tested for unidimensionality. The Bartlett test and the KMO measure of sampling adequacy indicate suitability for PCA. While all the items load above 0.32, the variance explained was only 32.65 per cent and Cronbach s Alpha was estimated at 0.57. Thus, a decision was made to refine the measure by investigating the items individual MSA (measures of sampling adequacy) (see Hair et al., 1995). Two items, the equipment/people focus and back/front office indicated individual MSA scores of below.50 (.460 and.476 respectively), thus, the measure was refined by excluding these two items. A second factor analysis was conducted with the remaining four items. The MSA for the individual items were all above 0.5 (see Hair et al., 1995). The details of the factor analysis undertaken are presented in Table 8. PLACE TABLE 8 HERE While results from the factor analysis revealed no difference in the reliability level between the first and second analysis, it is worth noting however, that the variance explained by the second analysis is higher (44.21 per cent) as compared to the first analysis (32.65 per cent). This indicates that the items had a higher explanatory power in measuring the construct for the sample in this study. Thus, four items namely, short/long contact, low/high customisation, low/high discretion and product/process focus were summated into a composite variable for subsequent analysis. 17

Cost Leadership and Differentiation Strategies Survey instruments utilizing Porter s (1980, 1985) competitive strategy have repeatedly been developed and refined in prior MCS studies (see, for example, Govindarajan, 1988; Miller, 1988; Kumar and Subramaniam, 1997; Chenhall and Langfield-Smith, 1998). In this study, the instrument to measure cost leadership and differentiation strategies was adopted and modified based on published items of Chenhall and Langfield-Smith (1998) and Kumar and Subramaniam (1997). To measure cost leadership strategy, respondents were asked to indicate on a 7-point scale, the degree of emphasis on the following activities: 1. Achieving lower cost of services than competitors 2. Making services/procedures more cost efficient 3. Improving the cost required for coordination of various services 4. Improving the utilization of available equipment, services and facilities High emphasis on these items indicated a higher emphasis on a cost leadership strategy. For the product differentiation strategy, respondents were asked to indicate on a 7-point scale, their degree of emphasis on the following activities: 1. Introducing new services/procedures quickly 2. Providing services that are distinct from that of competitors 3. Offering a broader range of services than the competitors 4. Improving the time it takes to provide services to customers 5. Providing high quality services 6. Customizing services to customers need 7. Providing after-sale service and support High emphasis on these items indicated a higher emphasis on a differentiation strategy. The items measuring cost leadership and differentiation were consequently tested for construct validity, utilizing the same analysis as undertaken in the previous measures developed in this study. PLACE TABLE 9 HERE The examination of cost leadership and differentiation item reveals that they were adequate for factor analysis. The reliability for these items was calculated at 0.81 and 0.77 respectively. Results of the factor analysis are summarized in Table 9. All items for the cost leadership and 18

differentiation strategies load above 0.32. Items for both strategy variables were summated for subsequent analysis. Organizational Lifecycle Stage The instrument to measure organizational lifecycle was developed based on Kazanjian and Drazin (1990) self-categorization measure. Kazanjian and Drazin s approach was seen as the most appropriate since the current study is adopting a contingency perspective, and the life cycle stages would be utilized as a contingent variable. However, some modifications had to be made as the instrument was developed for technology-based new ventures. The description offered at each stage of the life cycle by Kazanjian and Drazin was compared with the Quinn and Cameron s (1983) summary of nine life cycle models. Quinn and Cameron s summary offers an extensive review of life cycle stages as it integrates nine life cycle models from various areas. The description is also more comprehensive. Consequently, three stages of organizational life cycle were measured, namely the formation, growth and mature stages. This study excluded the decline stage although with this exclusion, the organizational life cycle would be incomplete. This notion is undertaken for two reasons; first, the stages from birth to maturity may be the only predictable ones (Quinn and Cameron, 1983), and second, the decline stage would not fit into the overall research purpose to predict organizational variables that could enhance organizational effectiveness. Stage 1 indicates formation, stage 2 is growth, and stage 3 is maturity. Respondents were asked to choose one box only that described their stage of development. A low number of firms self-classified as stage 1 (5 firms) led to the combining of stage 1 and stage 2 firms into one category, namely the growth stage. Thus, only two stages were analysed in the subsequent analysis, the growth and mature stages. The two stages now resemble the dichotomous distinction previously used by Dodge, Fullerton and Robbins (1994). An ANOVA was conducted for two contextual variables namely, organization age and size (number of equivalent employee) to provide some validity for this measure. Prior studies of life cycle stages have used age and size as common life cycle stage measures (Kazanjian, 1988; Hanks et al., 1994; Moores and Yuen, 2001). As evidenced in those studies, it was also expected that the age and size of the growth firms in this study would be lower than the mature firms. The result indicates significant differences between the size (F=2.074, p<.10, one-tailed) and the age (F=2.603, p<.05, one-tailed) with the mean age and size of growth firms being lower than the mature firms. As the results support the expectations, this provides some confidence for using the developed self-categorization measure. 19

RESULTS Table 10 presents the descriptive statistics for all variables and Table 11 displays the correlation matrix. Examination of the correlation matrix indicates that there are no high correlations (greater than.70) among the contingent variables, thus there is strong reason to believe that multicollinearity is unlikely to be a problem (see Tabachnick and Fidell, 2001, p. 84). PLACE TABLES 10 AND 11 HERE Hypotheses 1, 2 and 3 The students t test was used to examine how the emphasis on MCS differed across different contingent situations as stated in H 1, H 2 and H 3. The comparison is made between two different groups. For example, comparing the MCS means between mass service and professional service firms, a t test was undertaken on two independent samples. A significant difference between the means of the MCS for these two groups would support the hypothesis. Table 12 contains the results for H 1, H 2 and H 3. H 1 predicted that mass service firms would place a greater emphasis on a more bureaucratic MCS compared to professional service firms. Given the continuous scale for the variable service process type, a median split was undertaken to classify firms within the mass service and professional service groups. Cases with a summated score of less than the median 4.75 were classified as mass service firms, while cases with a score more than or equal to the median 4.75 were professional service firms. As shown in Table 12, there is a significant difference (t = 3.313, p.01) between the MCS means for these two groups, with the mean for mass services lower than the mean for professional services. This result indicates support for H 1. PLACE TABLE 12 HERE H 2 proposed that a more bureaucratic form of MCS would be emphasized by cost leaders compared to differentiators. As business strategy was measured on two separate dimensions of cost leadership and differentiation, the measurement will capture groups pursuing (1) high on both strategies, (2) low on both strategies, and (3) low on one strategy, but high on the other. Acknowledging this, a median split was undertaken to separate high cost leaders from low cost leaders, and high differentiators from low differentiators. The cost leaders are those firms that place high emphasis on a cost leadership strategy (median 5.25) but low emphasis on a differentiation strategy (median<5.43). The differentiators are those firms that place high emphasis on a differentiation strategy (median 5.43) and low emphasis on a cost leadership strategy (median<5.25). As shown in Table 12, the mean of MCS for the cost leaders is 20