1
Amsterdam Business School
Moderating effect of employee age and experience on the relation between task complexity and control tightness in professional service firms
Name: Aiperi Ergeshova
Thesis supervisor: Sander van Triest Student number: 10824693
Date: 12559
MSc Accountancy and Control, specialization in Accountancy Faculty of Economics and Business, University of Amsterdam
2 Statement of Originality
This document is written by student Aiperi Ergeshova who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
This master thesis aims to investigate the composition and design of the management control system in the professional service firms. Despite the growing influence of the professional service firms on the work economy these firms was not studied extensively, and existing literature regarding PSFs is embryonic and stays relatively general and industry unspecific in their elaboration (Malmi and Brown, 2008).
Professional service firms classified as knowledge intense firms Von Nordenflycht (2010) and therefore it is a vital to employ and manage highly skilled employees in an efficient and effective way in order to increase market share and the competitive advantage of the company. Therefore, the main research question of this study is what kind of control system professionals experience in their organization. Based on the contingency theory, this study assumes that task complexity is one of the factors that influence MCS and therefore, the first research question aims to investigate the relation between task complexity and control system tightness. Moreover, this study analyzes the moderating effect of the employee age and job tenure on the relation between
The results of the study support the assumption that task complexity does have an influence on the control system tightness. However, I could find support only regarding the behavioral control tightness, and no justification of the hypothesis on result control tightness. As for the moderating effect, the outcomes of the analysis shows positive interaction effect of employee age on the relation between task complexity and action control tightness, however, the results did not support the hypothesis regarding the job tenure.
1
Table of content
1. Introduction ... 2
2. Theoretical Background ... 4
2.1 Professional Service Firms... 4
2.2 Management control system ... 6
2.2.1 Looseness and tightness of MCS ... 7
2.2.2 Explicit and Implicit control tightness ... 10
2.3 Task complexity ... 10
2.3.1 Background on task ... 10
2.3.2 Characteristics of task complexity ... 11
2.1 Employee age ... 12
2.2 Employee experience ... 13
3. Hypothesis development ... 14
4. Research methodology ... 17
4.1 Data collection and research sample ... 17
4.2 Operationalization and measurement of the constructs ... 23
4.2.1 Task complexity construct ... 23
4.2.2 Result control tightness construct ... 24
4.2.3 Action control tightness construct ... 25
4.2.4 Control variables ... 25
4.2.4.1 Organization size ... 25
4.2.4.2 Education level ... 26
4.2.4.3 Gender ... 26
5. Preliminary analysis ... 28
6. Data analysis procedures and equations ... 31
7. Results ... 32
7.1 Additional test ... 33
8. Discussion and conclusion ... 31
2
1. Introduction
Last decades professional service industry expanded rapidly and became more pronounced in the economies of the world over. Professional service firms (hereafter PSFs) represent growing share of both the population and GDP. Moreover, in today’s integrated global environment PSFs became an increasingly important driver of economic activity in the knowledge economy (Kox 2002, Kox et al. 2003, Den Hertog 2000). Accounting, architecture, engineering, legal, health care and management consulting are typical examples of PSFs. Given the importance of the PSFs in today’s economy, PSFs have not been studied extensively, and existing literature regarding PSFs is embryonic and stays relatively general and industry unspecific in their elaboration (Malmi and Brown, 2008).
It is generally accepted that service firms are different from manufacturing companies in at least four distinctive features. First of all, service firms provide intangible service, whereas manufacturing companies provide tangible goods. Secondly, these companies usually do not hold inventory as the production companies do. Moreover, service companies do not produce service unless the customer wants it. And finally, these companies recruit employee with specific knowledge. Combining all these characteristics an initial viewpoint states that PSFs are firms, whose primary assets are highly educated (professional) workforce and whose output is intangible service encoded with sophisticated knowledge (Nonaka and Teece, 2001; Greenwood et al., 2005). This approach parallels with Von Nordenflycht (2010) who identifies three key features of PSFs such as knowledge intensity, low capital intensity, and professionalized workforce that distinguish PSFs from manufacturing firms. Therefore, as Lowendahl et al. (2001) state knowledge is the strategic resource of PSFs and employees are engines for storing this knowledge. Accordingly, human capital is considered as the most valuable assets of PSFs and the designing management control system with appropriate combination of control plays a vital role in getting a competitive advantage and increasing market share. However, this aspect of the PSFs did not receive much attention in the prior studies and findings of the previous studies are controversial.
Contingency theory advocates that all companies differ from each other and there is no universally accepted best MCS. Thus, Wiio and Goldhaber (1993) mention that there are external and internal constraints that may have an influence on operation companies and therefore designing and implementing MCS requires consideration of various factors like economic, technological, legal, socio-politico-cultural and environmental contingencies as well as structural contingencies, output, demographic. Based on this theory I assume that one of those factors that might influence desing and implementation of the MCS in PSFs is task complexity. Jarowski and Kohli (1993) defined task complexity as "the extent of predictability and variety in the activities to be performed for a given
3 position". In this study I hypnotize that task complexity does have an influence on the MCS, and I propose that task complexity is one of the important variables, according to which management control system is designed. To address this assumption, this research aims to examine the relation between task complexity and management control system tightness. This assumption implies the discovery of the fist research question which is:
RQ1: How does task complexity impact control tightness in professional service firms?
Knowledge is considered as a core resource for professional service companies. However, knowledge is subject to change and innovation. It is innovative concepts and new services that make a professional service company a market leader (Kaiser and Ringsletter, 2011). In most cases innovation and changes in the economic environment have an influence on the nature of the task by making it more complicated and challenging and therefore, it can be concluded that change leads to more complex tasks. According to Campbell (1988), there are three different treatments for the complexity. Fistly task classified as complex due to the psychological perception of an employee. Another treatment is as the relation between task and personal characteristics, and finally as functions of objective task characteristics. Since change and innovation influences all these three categories this implies that in order to get competitive advantage companies and particularly employees have to respond to these changes by adopting and learning. However, employees may react differently to the changes and work complexes, and there is a need for a different control system for better performance. Usually, these differences in attitude to the changes are affected by employee age and his (her) work experience. Based on this assumption I would like to investigate the moderating effect of employee age and expertise on control tightness. So, second research question is:
RQ2: What are the effect of employee age and work experience on the relation between task complexity and control tightness?
This study contributes to the existing literature by analyzing the design of management control system in professional service firms. It should be clarified that despite the importance of such variable as task complexity in developing control system in PSFs, no research was conducted previously analyzing this relationship. Additionally, optimistic view of contribution is applicability of results in practice by PSFs.
The remainder of this paper is structured as follows: Section 2 provides literature overview and section 3 focuses on hypothesis development. Further Section 4 shows research methodology and
4 the results of the study are presented in Section 5 followed by conclusion and discussion in Section 6. And finally, the paper concludes with necessary literature references and appendixes.
2. Theoretical Background
There is an ambiguity in defining the PSFs mainly caused by the existence of uncertainty in defining the term professional. So this section aims to analyze existing literature and determine what PSFs are and provide their key features. Moreover, this section seeks to provide findings of previous researchers regarding task complexity, management control system, employee age, and experience. These explanations will help to understand constraints and provide reasoning to my research question. Additionally, this section highlights motives for hypothesis development.
2.1 Professional Service Firms
Over the past three decades, the PSFs became one of the most rapidly growing, profitable, and significant sector of the global economy. The emerging literature states that PSFs are different from manufacturing firms; therefore, PSFs have been of growing interest to researchers. Nevertheless, the organizational structure and characteristics of PSFs can be described from various aspects.
As a starting point, I will analyze typology of PSFs based on the Vaassen et al. (2009) framework. Thus, Vaassen et al. (2009) in his works developed several types of organization typologies based on the source of income and internal control system. Authors distinguish among two major types of organizations such as organizations with the dominant flow of goods, which usually refers to trade companies and organizations without the dominant flow of goods, which mainly composed of service firms. Furthermore, organizations with a limited flow of goods are classified into three categories such as service organization with a limited flow of goods, an organization that put space and electronic capacity at their customers’ disposal and companies that put knowledge and skills at their customers’ disposal. Based on these framework PSFs are classified as organizations with a limited flow of goods that put knowledge and skills at their customers’ disposal. These organizations generate revenue by selling/disposition of man hours and intellectual property or selling financial products. This classification is also supported by Winch and Schneider (1993) who define service firms as organizations that deploy the assets in a such a way sell capacity to produce rather finished products. Therefore, PSFs in comparison with another type of companies are pure knowledge intensive firms and they face a distinctive environment that demands different theories of management (von Nordenflycht, 2010).
Existing literature is controversial, and there is little agreement between researchers in identifying definition and characteristics of PSFs. Lowendahl (2001) characterizes PSFs as the
5 organizations that employs high proportion of highly education and dependent on the abilities of employees to attract, mobilize and develop knowledge to create value for the clients. Another but yet similar explanation is provided by von Nordenflycht (2010), which states that PSFs are organizations whose primary source of income are members of the professions. However, PSFs differ from each other by their occupations, target group and kind of service provided therefore there is a need for a more comprehensive definition. Thus, von Nordenflycht (2010) identifies and focuses on three key features of the PSFs, which are knowledge intensity, low capital intensity,
and professionalized workforce.
Knowledge intensity indicates that production of a firm’s output relies on a substantial body of sophisticated knowledge (Starbuck, 1992; Winch and Schneider, 1993) and company mainly use intellectually skilled employees at every level of organization, not only on the higher level of the organizational hierarchy (von Nordenflycht, 2010). Low capital intensity means that company does not intensively exploit assets other than human capital, and there is no need for significant investments in tangible assets like equipment, inventory, etc. Moreover, low capital intensity increases the importance of the professionals since human capital is one of the most valuable assets of the firm and the only asset that defines the quality of the service provided. Another characteristic of the PSFs is the professionalized workforce which according to von Nordenflycht (2010) is the presence of two institutional features of professionalization such as ideology and self-regulation. Here professional thinking can be explained as explicit ethical codes enforced by professional associations and in internalized preferences often developed during professional training (Leicht and Lyman, 2006). Self-regulation, on the other hand, means that professionals have substantial control over the practice of the occupation (von Nordenflycht, 2010)
Among these three characteristics, knowledge intensity is the term that is used in the primary throughout this paper. Jensen and Meckling (1992) in their studies distinguish among specific and general knowledge. Here specific knowledge defined as the one that costly to transfer while general knowledge is inexpensive to transmit. Professionals gain knowledge through formal education and training during the job. This education and training usually provide a high level of articulable knowledge in the field of specialty. Therefore, the human element has grown in importance because education has become a critical prerequisite for have a competitive advantage, particularly in the new economy landscape (Grant,1996).
Since human capital is argued as an essential resource for most firms (Pfeffer, 1994) implementation of appropriate MCS is a crucial element in the success of every business. Yet, despite increasing interest of the PSFs, previous studies mostly focused on the particular industry
6 and very broad in interpretation. Accordingly, little work was conducted to analyze MCS in PSFs even though MCS in PSFs are perceived to be more complicated, due to the fact in addition to importance of attracting “righ” employees PSFs have to deal with knowledge integration of employees to carry out activities in a conditions of knowledge asymmetry, uncertainty and problems (Ditillo, 2004). So, the primary aim of this study is to investigate the management control system tightness in PSFs by applying moderations like employee age and experience.
2.2 Management control system
Existing literature provides controversial definitions and descriptions of the term management control system. However, despite different interpretations, all researchers highlight importance and necessity for appropriate MCS. Accordingly, Bloom and Van Reenen (2010) state that firms that have excellent management practice tend to have better performance in comparison with those who have a weak control system. Some authors define MCS as a broad concept that encompasses management accounting system (Chenhall, 2003), While others note, that it is more general idea of control that includes factors such as strategic development, strategic control, and learning processes, all of which are typically beyond the scope of management accounting (Merchant and Otley, 2007). The narrowest definition provided by Merchant and Van der Stede (2007) and defined as a process by which managers ensure that resources are obtained and used effectively to achieve company objectives. While analyzing MCS, I would like to note the direct link between MCS and general/specific knowledge. Thus, MCS is applied with the aim to collect and use knowledge/information to evaluate organizational resources that influence the performance of the organization.
Contingency theory states that there is no universally best management system applies to all kind of organizations and situations. Several factors may affect the structure of the management control system. Amongst factors that have a possible impact on MCS is the strategy of the firm, external environment, technology, organizational structure, firm size, and culture. PSFs should consider all these factors in designing and implementing control system to have pronounced and effective MCS. In addition to contingency theory, researchers distinguish between several kinds of control techniques that are applied in MCS. In my research, I will be using control types provided by Merchant and Van der Stede (2007). Authors distinguish among four types of management control which are: result controls, action (behavioral) controls, cultural controls, and personnel controls.
Result controls define performance measures and hold employees accountable for the delivered results, by rewarding or punishing them. They also facilitate empowerment as actions are not constrained and provide autonomy for employees. On the other hand, action control focus on
7 employee behavior to ensure that employee performs actions beneficial to the organization. Under this type of control decisions/actions are observable and monitored by a supervisor. Cultural controls refer to shared norms and values, and social pressure exerted by groups on individuals while personnel controls were implemented in the selection of employees and training. Similar conceptualization can be found in the works of Ouchi (1980) as market, bureaucratic, clan and social controls. It should be noted that according to the conceptual framework (Figure 1) provided in the works of Merchant and Van der Stede (2012) action control requires high level knowledge of actions required and high ability to measure the outcomes, result controls on the other hand may or may not require actions necessary and requires high level of ability to measure the results on performance dimension. Personnel and cultural control requires a small degree of knowledge of the desired actions and are not able to measure the results of the critical performance dimensions.
Figure 1 - Merchant and Van der Stede conceptual framework (p.55)
These four controls are categorized by Jaworski (1993) into two broad classes of controls – formal (soft) and informal (hard). Formal controls are defined as management initiated mechanisms that influence the probability that marketing personnel will behave in ways that support the stated marketing objectives. Informal controls, on the other hand, are unwritten, typically worker-initiated mechanisms designed to influence the behavior of marketing personnel (Jaworski, 1988). Therefore, based on the definitions result and behavioral controls are categorized as formal controls, whereas personnel, and cultural controls are classified as informal controls.
2.2.1 Looseness and tightness of MCS
The benefit of having MCS can be derived from the increase in the probability that company objectives are achieved. Merchant and Van der Stede (2012) argues that this benefit can be described as tightness or looseness of MCS. Existing literature does not provide precise definition and
8 operationalization of these terms and the most appropriate description, in my opinion, is provided by Merchant (1985b) who defines tight control as a limitation to the degree of freedom of the individual who is being controlled, which is, in other words, can be described as the level of flexibility in the control system. In another work Merchant (1985a) describes tight control if it provides a high degree of certainty that employees act in a required manner. These two definitions, in my opinion, are the best description of the control tightness; therefore, in my study, I will be using these definitions.
Tightness can be created in two methods. The first way is by increasing the extent or scope of the MCS, which means creating more controls, more rules, and more procedures. The second way is extending the level of tolerance for deviations from the MCS, which can be accomplished by minimizing the difference in scope between the actions defined by the control system and those deemed acceptable within the organization. In this study, two types control are analyzed, which are action and result in control tightness. In defining control tightness, I mostly refer to the work of the
2.2.1.1 Result control tightness
Marchant and Van der Stede (2012) in their studies of MCS pointed out three ultimate requirements for result control tightness which are the identification of the desired results, adequacy of the performance measures, and incentives.
The fist prerequisite is the degree to which controls aligned with the objectives of the organization. Thus, authors advocate that not only controls but also control measurement dimensions should reflect company real goals and managers, as well as employees, should comprehend these objectives. Moreover, controls should be precise, which means that employees should have targets described in the specific terms, ideally expressed in the quantitatively. The absence of specific description of the goals and objective may result in a decrease of job motivation and reduce the overall performance of employees. Moreover, it is crucial for these objectives to be communicated to professionals efficiently and employee should believe that goals are achievable.
The second requirement for result control tightness is the adequacy of the performance measures, which Merchant and Van der Stede (2012) point that controls have to rely on “measures that are precise, objective, timely and understandable”. And the final requirement is the availability of the reward that is directly or indirectly linked to the achievement of the desired results. This view parallels with interpretation of Lerner and Wanat (1983), who argue tight control requires that decision rules are precisely defined, while in loose control environment precise rules may exist but these rules are subject to change in accordance with particular situation and employees.
9 2.2.1.2 Action control tightness
The design and types of behavioral control vary depending on the organization size, type, and targets. Merchant and Van der Stede (2012) suggest that tight action control can be achieved only with employee’s consistent engagement in critical actions that are vital to the success of the company. However, another view of control tightness is presented in work of Hopwood (1974), who declares in tight control employees’ participation in setting objectives is low. Merchant and Van der Stede (2012) distinguish among types of action control like behavioral constraints, preaction review, and action accountability. Behavioral constraints which also may be called as administrative constraint is the primary type of action control and might the one that is most widely used control. It may come in many forms such as restricting decision authority and separating duties between employees. The first one refers to the case when higher level personnel make a more reliable decision than the lower level employees. The second one, on the other hand, is designed to restrict the right of employees in performing their duties to reduce the risk of the harmful activity. The second degree is also widely used in the internal control literature and called as segregation of duties. These kinds of controls are accepted as one of the most efficient tools and the cornerstone of the internal control system.
Preaction review can affect control tightness if reviews are conducted consistently and frequently and carried out by experienced, knowledgeable supervisor/reviewer. Merchant and Van der Stede (2012) in their book argue that these controls are mostly applied in the areas involving large resource allocation and usually expressed by frequent control of business plans and request for capital allocations.
And the last degree of action control is action accountability, which somehow may be similar to result control. According to authors, action accountability control depends on desirable actions, the effectiveness of the action tracking system and the reward and punishment system. The essential condition for action control tightness is congruence, preciseness, completeness and communication of the desired actions. These criteria are similar to the result controls prerequisites. Therefore, I will not go into details to explain them. Another tool for action control tightness is the effectiveness of action tracking system. It can be achieved by immediate notice of employees’ behavior to limit undesirable activities.
Additionally, tight action control is also associated making reward and punishment to affect employees more significantly. It can be achieved by making a direct link between reward and punishment and employees’ behavior.
10 2.2.2 Explicit and Implicit control tightness
Another realm for classification of the control tightness is depending on explicitness and implicitness of controls. Increased tightness achieved through control selection, definition, and completeness, calls this explicit tightness. On the other hand, tightness formed by decreasing the level of tolerance for deviations from the MCS is named as implicit tightness.
2.3 Task complexity
The concept of task complexity has been considered in the previous literature as an essential element in at least three fields: task and design literature, information making and decision making literature and goal setting research literature (Campbell, 1988). Additionally, importance of task complexity is outlined in the works of Ouchi and Maguire (1975), who suggest that task complexity influences the system of control in use. Therefore, it can be concluded based on the existing literature, task and task complexity play a significant role in goal setting, planning, motivation and strategy development process of every organization and especially of PSFs due to nature of economic value drivers of these companies. However, surprisingly most of the studies covering task complexity were conducted in the late 19th and even though the economic and working environment dramatically changed, recent studies of task complexity are limited. Therefore, there is a need for additional studies to examine the role of task complexity in MCS. As a starting point I would like to describe the notion of task itself and then continue with the characteristics that make task complex.
2.3.1 Background on task
Previous literature is not uniform on providing a definition of task and task complexity. First of all in this section, I want to provide theoretical underpinning and characteristics of the term task itself. Wood (1986) in his paper argues that there is no adequate theoretical model to describe task and how they differ from each other. Therefore, he exploits empirical and theoretical approaches to analyzing characteristics of the task. Here, the empirical approach is based on the individual perception of the task. Whereas, theoretical approach builds on the work of Hackman (1969) and distinguishes among four conceptualization of task which is task qua task, a task as a behavioral requirement, a task as a behavioral description, a task as ability requirements.
Task as behavioral description approach defines task by whatever the group members actually do in order to achieve performance, whereas under task as behavioral requirement approach task is defined by the actions should be taken with a view to performing task, and task as ability differentiates based on the skills required to complete task, and finally, task as qua is defined “as the pattern of stimuli impinging on the individual” (Hackman, 1969). Even though author argues that it is challenging to separate the effect of task characteristics from different effects, task qua task is the
11 only approach that adequately characterized task and only flexible framework with feasible task constructs (Woods, 1986).
Additionally, Wood (1986) suggests that task contains three essential components like products, required acts and information cues. Here products are defined as the measurable result of actions and every task identified and deviated from each other depending on the product. Acts, on the other hand, are the required state of actions that are needed to the creation of the defined products. Acts can take a very particular form as well as very abstract and complex actions with the defined purposes. Information cues are the parts of information upon which professional bases his judgment and which is required to perform a task (Wood, 1986).
Subsequently, Wood (1986) defines task complexity as the relationship between two components like required acts and information cues. This framework is in compliance with the definition provided by Jaworski et al. (1993) who defines task complexity as the degree of predictability and diversity of the activities to be executed to perform the task. This approach complies with the definition outlined by March and Simon (1958) who identified several objective task characteristics that contribute to task complexity. Thus, authors argue that complex task is characterized by unknown or uncertain alternatives or consequences of action, furthermore task are characterized by inexact or unknown means-end connections and finally by the existence of some subtasks, which may or may not factored into nearly independent parts (March and Simon, 1958). Another interpretation of task complexity is provided by Cambell (1988) where task complexity is defined as any objective task characteristics that imply an increase in information diversity and load, or rate of information. The definition of task complexity provided by Jaworski et al. (1993) is a basis of this study and used primarily throughout the paper.
2.3.2 Characteristics of task complexity
Cambell (1988) questioned properties that make task complex and propose that complexity can be threatened in three different dimensions, which are: psychological experience, the interaction between task and personal characteristics and function of objective task characteristics. Moreover, the author distinguishes between task complexity as subjective to the particular employee or as objective to the whole company. The subjectivity of the task depends on the skills, education of the employee performing the task and purpose.
Moreover, Jaworski (1993) identified four dimensions of task complexity such as Interdependence, routiness, “learn-job time” and completeness. The first dimension refers to the degree to which one departments’ performance depends on performance of another department.
12 This dimension increases the task complexity in a way that it creates the problem of control when task requires joint effort of separate units (Dalton 1971; Otley 1980). Furthermore, Baulmer (1971) suggests that formal controls are insufficient when there is a high interpedendencies between department, therefore he proposes use of clan system.
The second view it the routiness which means tractability and ease of measurability of activities to be performed on the job (Jaworksi, 1993). This definition paralles definition provided by Rico et al. (2008) who advocate that high task rouitiness represent situation where task can be performed by applying standardized procedures, non routine task on the other hand are associated with more unique acts and frequent changes of the requirements. In fact, considering the necessacity for different alternative actios non routiness of task complexity influences the successful completion of the task (Diefendorff et al., 2006).
Learn-job time dimension refers to the time required to learn the job. Some task require simple action to be taken and therefore easy to learn, while others may require complex an comprehensive decision making therefore require more effort (Jaworski, 1993). And finally the last dimension decribed by Jaworski (1993) is the completeness. Here completeness is described based on the defintion provided by Anderson and Chambers (1985) and defined as the extent to which the evaluation system adequately captures the full range of job responsibilities and activities. Dimension such as routiness, learn-job time and completeness is used in this study to support the hypotheses.
2.1 Employee age
According to the report of the United Nations, population aging is occurring in almost every country since “baby boomers” are turning their sixties. These demographic changes have a significant effect on economics like increasing expenditures on health care, pension funds and retirement payments, long-term care services. Researchers state that population aging has an effect on workforce composition by increasing the employment rates among older workers (Kooij et al. 2011; Robson and Hansson, 2007). Obviously, people are subject to change physically and psychologically with age and these changes usually have a direct effect on the work attitude of the employees. Therefore, demographic changes require companies to have separate and unique MCS that will motivate older employee to behave in the best way since certain things older people do better than young people and vice versa. Subsequent research states that older people in comparison to younger people are less vital and open (Roberts et al., 2006). On the other hand, other studies state that seniors are more self-controlled, tolerant, modest and conscientious (Kanfer and Acherman, 2004; Warr, 2001).
13 The concept of aging has been studied by researchers from different disciplines incorporating psychology, political, economic and behavioral studies. So, first of all, we have to define how age is described in the existing literature and what kind of age characteristics exists. Age is a multidimensional aspect which influences performance, decision-making process, behavior and productiveness of an individual. In this study age is going to be analyzed in two dimensions. Thus, Cleveland and Lim (2007) in their studies make a distinction between persons’ real age and persons’ perception of age. First, one is referred to chronological aging that is expressed in the number of years a person lived or perceived age which refers to how old or young individuals see themselves (Steitz and McClay, 1988). The second dimension is the mental age which is expressed through job experience. These two aspects of age usually are not the same, and two employees with the same chronological age may vary in performance, attitude and behavior due to the influence of many factors like education, work experience, living conditions, etc.
Results of studies about the relationship between age, experience, knowledge and performance are not consistent. For instance, Warr (2001) in his studies shows a positive correlation between age and knowledge. While Ackerman (2000) argued that experience should be relevant to the field of assessment, since every field has its specific knowledge. However, despite the existence of a positive relationship between experience, age, and knowledge, meta-analysis shows weak support for age-task performance rating relationship (Avolio, Waldman, and McDaniel, 1990). Additionally, Sterns and McDaniel (1994) in their studies proved the weak positive correlation between age and performance. The existence of controversies among studies of the relation between age, experience and performance increases the interest of this topic. Therefore, in this study, both chronological age and work experience of the employee will be analyzed separately.
Employee age also has been considered by the working memory paradigm notion. Working memory is defined as the system for temporarily storing and managing the information to perform complex tasks and requires immediate processing of information. Thus, a study by Welford (1958) shows that older people perform poorly on perpetual motor work, and the possible explanation of it is an age-related decrease in the short-term memory; moreover, older people are more susceptible to interference from the performance of other activities. Another study proved that seniors face problem due to the reduction of working memory capacity (Cohen, 1981). And these aspects may differ depending on the task complexity.
2.2 Employee experience
One of the factors that influence work performance is the existence of necessary knowledge, skills and work experience to perform job tasks. Thus experience contains work-related experience,
14 skills, and motivation which in fact affect employee performance. Moreover, work experience in important in many human resource functions like selection (Ash and Lavine, 1985), training and career development (Campion, Cherasldn, and Stevens, 1994; McCall, Lombardo, and Morrison, 1988). Therefore, it has been studied extensively previously. At this stage, it is important to understand the meaning and measurement of work experience.
Psychological literature defines job experience as job-related knowledge gained over time (Fiedler, 1970; MeCall et al., 1988). For instance, it is assumed that if the employee performs a particular job, then professional gains knowledge in the field of the task performed and therefore become more experienced. Thus, existing literature describes the influence of the work experience on job performance as well as the relation between mental ability and job performance (Schmidt, Hunter, and Outerbridge, 1986). However, some studies show a weak relationship between education/experience and performance (Schmidt and McDaniel, 1985).
However, the literature is not consistent in measuring the experience. Thus, Ford et al. (1992) in their studies of U.S Air Force identifies their models to measure work experience. First is the number of the task performed which is breath. The second method is the number of times tasks a performed which is called as activity level, and the third is the difficulty of the task which is task type. Therefore, in measuring the job performance, it should be distinguished from measurement mode and the level of specificity (Ouinones, Ford and Teachout, 1995). Quinones et al. (1995) in their study of the job experience propose time-based and amount-based measure. Here time-based means job tenure, while amount refers to activity level which was mentioned in the work of Ford et al. (1992). Here should be noted that job tenure might also be different depending on the prestige, difficulty, criticality or contribution of the task performed (Ouinones, Ford and Teachout, 1995).
3. Hypothesis development
Management control is a vital function in every organization. Failures in implementing appropriate management control system can result in substantial financial losses and may have an adverse influence on the reputation of the firm. By selecting necessary and adequate MCS, the company can improve the use of human capital and gain higher customer satisfaction. Therefore, it is very critical for the enterprise to have efficient and effective MCS that will work in the best interest of the company. As it was mentioned before PSFs are different from other kinds of organizations in many ways, so there is a need for the particular type of MCS that is capable of creating and utilizing knowledge in the best interest of the company. Moreover, MCS became even more vital due to the human capital intensity of PSFs, since it is people who make things happen.
15 Additionally, following contingency theory, the company should consider factors that are important in designing MCS like strategy, specific legal, cultural and organizational characteristics of professional service firms when implementing MCS (Kaiser and Ringsletter, 2010).
Despite the importance of the MCS of PSFs, previous studies covering control system of these companies are embryonic. For instance, an issue that is not addressed previously and which this study aims to analyze is how task complexity influences the management control system in professional service firms. To be more specific, this study seeks to investigate how task complexity affects the design of MCS in PSFs considering MCS framework proposed by Merchant and Van der Stede (2007). From the literature review section, it can be concluded that task complexity influences MCS in many ways, and it should be considered as an important factor in designing MCS. Existing literature suggests that complex task and a challenging goal might result in improved task performance due to the effects of both cognitive and motivational processes (Campbell and Gingrich, 1986; Earley, Wojnaroski, and Prest, 1987; Terborg and Miller, 1978).
Another issue that should be considered in designing MCS in PSFs is that they are categorized as full knowledge intense firms. Knowledge intensity of PSFs leads to a high reliance on the human capital and requires a particular type of MCS that differs from manufacturing companies (von Nordenflycht, 2010). Therefore, one of the main questions that should be answered is what kind of controls and to what extent should be implemented to have an effective and efficient MCS. It might be assumed that employees of PSFs require more autonomy and fewer behavioral controls in performing their responsibilities since they already have experience and enough knowledge in the job field. Moreover, as Merchant (1985b) stated in his study control should be feasible meaning that if action controls are applied then, management must know desirable actions must be taken to complete the task. However, it is not always possible to determine steps to be taken in performing complex tasks. On the other hand, it might be argued that task complexity, controversially, may lead to more tight action control because employees, when faced with a complex task, need detailed guidelines and controls to have better results. This assumption leads to the first two hypotheses of this research which are:
H1: Task complexity is negatively associated with tight action control H2: Task complexity is positively related to tight result control
Previous studies state that knowledge is considered to be core resource of professional service firms. However, type of knowledge and demand for it may change over time due to development and innovation. Mostly these changes may lead to more complex and arduous task to be performed
16 by employees and therefore PSFs have to respond to these changes to gain competitive advantage. Let’s consider the example of the accounting firms. Recently we have witnessed an increase in demand and reliance on IT infrastructure. Decades ago nobody could imagine that IT will play a crucial role in the success of the firm. Today every organization no matter how small or big requires IT system that deals with automatization of the business processes, communication between divisions and provision security over access and storage of data. Therefore, to provide reasonable assurance over company operations accounting firm should have employees who understand and have knowledge of IT infrastructure and consider IT and another risk during their works no matter how much experience he/she has in the accounting sphere. Thus, there is a need for an experienced employee at their mature years to adapt quickly to the changes and a new generation of employees who was grown in the new IT era.
However, employees may respond differently to developments and innovations. Some may accept and adapt to these changes quickly while other may refuse to adopt and change their perceived practice. Mostly, this difference in perception is due to employee age. Thus, it may be argued that older employees are seen to be more conservative and have little ability to adapt to development and innovations. Indeed, the result of a study of Sharit and Czaja (1999) shows those young participants in the research were able to balance a larger number of transactions than older employees. Therefore, it may we assumed that older employees, despite their work experience, need tight action control when faced with the complex task. On the other hand, considerable research in manufacturing firms indicated that job performance does not decline significantly with age (Salthouse and Mourer, 1996). Additionally, Sharit et al. (2004) investigated that older employee’s performance on telecommunication task like some emails responded correctly is higher than of the younger employees. This inconsistency in the existing literature motivated me to the next hypothesis which is:
H3: As the level of task complexity increases mature employees experience more tight action control than younger employees and less tight result control
However, despite the fact that some literature argues that the working capacity of the mature professionals diminishes, greater knowledge of the older employee may compensate for age-related declines in information processing (Warr, 2001). Moreover, some studies proposed decline of thriving at work at an older age due to a decrease in succeeding, job motivation and openness to changes are drop. Based on these findings it is assumed that people with difference age and work experience and the same task may require different types of the MCS. However, according to the literature review, age and experience are not always correlated; therefore, there is a need for separate
17 analysis of the effect of work experience and employee age. Therefore with the aim to investigate whether employee age and job experience have the same moderating on the relationship between task complexity and the control tightness I have developed next hypothesis which is:
H4: As the level of task complexity increases employees with more job tenure experience more tight action control and less tight result control
Figure 3.1 – Characteristics of constructing relationship
4. Research methodology
The study is conducted in the context of a project initiated by Faculty of Economics and Business (hereafter FEB) of the University of Amsterdam (hereafter UvA). The primary purpose of the project is to analyze the MCS in PSFs. This section describes data collection and sample selection processes, conceptualization and operationalization of all constructs that have been identified as determinants for study.
4.1 Data collection and research sample
With the ultimate purpose to investigate the relationship between task complexity, control tightness and employee age/experience a survey has been elaborated. Because this study is, the part of the UvA survey project, the template of the questionnaire was provided by FEB of UvA. Participating students had to acquire at least seven completed surveys. Survey contains 39 questions that are designed to analyze the coordination, motivation and performance measurement of employees.
18 Target respondents of the research are mid-level employees of medium to large professional service firms. Respondents must have at least three years of working experience in the fields and should be not less than 50 employees in the company. The reason why limited professional experience is that in the later stage employees are no longer in the learning phase and the motivation may be different from the early learning phase. As a result of the database analysis, it was revealed that 30 participants had less than three years of work experience. However, this is for the benefit of the research because it provides a better base for the analysis and comparison. Therefore, I decided to include these respondents in the sample and consider them in the further analysis. Another requirement is that respondent should not be the owner/partner or in any other position to exercise power over design and implementation of the MCS. The primary purpose of this requirement is that the study aims to analyze the effect on the MCS on the professionals. Additionally, this study is not limited to particular industry or country; therefore employees from a various kind of PSFs from all over the world were invited to fill in the survey. The advantage of it is that we will be able to compare the management control system in various industries and across different countries. With the joint effort, 371 employees participated in our study. Table 4.1 and Figure 4.1 provide overview of statistical information regarding industry, age, gender and experience of respondents:
Table 4.1 Respondent characteristics
Age Min / Max / Mean 19 / 63 / 34.46
Work experience in the current
field Min / Max / Mean less than 1/ More than 10 / 6.94
Occupation industry % Accounting 23.7
% Advertising 0.3
% Architecture 0.6
% Consulting Engineering, IT/Technology, and Management 19.1
% Engineering 2.8
% Financial advising 3.1
% Graphic design 0.3
% Investment Banking 1.9
% Investment management (hedge funds, VC,mutual funds) 0.8
% Law 3.6
% Marketing/public relations 4.2
% Media production (film, TV, music) 0.6
% Medicine/Physician practices 7.5
% Pharmaceutical 0.8
21
% Real Estate 1.7
% Recruiting - executive 1.7
% Research/RandD 1.7
% Risk management services 2.8
% Software development 1.4 % Other 19.5 Gender % Male 60.9 % Female 39.1
Education % Bachelor level or lower 36.8
% Master degree 52.9
% Ph.D. or another professional doctorate program 10.2
Figure 4.1 Age and gender distribution
To assess the quality and effectiveness of the survey, two separate pre-tests were conducted. In particular, the first pre-test was designed to evaluate the quality of the items used to measure the variables for the types of management controls. The first sheet of paper contained the construct definitions for the eight control constructs (implicit/explicit - results/behavior/cultural/personal control) and the second sheet of paper included 52 statements designed to test those constructs. Subjects were asked to match each statement to the definition they felt it most closely resembled. Second pre-test, on the other hand, was designed to assess the quality of the survey as a whole. An additional 20 subjects were asked to view the questionnaire online and answer a series of questions regarding the content, clarity and appearance of the survey as well as the amount of time required to
19-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 61-65
Age and gender distribution
23 complete the survey. Thirteen subjects provided written answers to the questions and the remaining (seven) provided answers by telephone. The comments given by the subjects resulted in only minor changes in wording and the inclusion of additional responses by a few multiple choice questions.
4.2 Operationalization and measurement of the constructs
The questionnaire provided by FEB of UvA is based on a review of the management control literature. The pre-developed questionnaire solicits of 39 broad to describe the variables. These items represent 14 constructs. From 14 constructs three of them will be used in this study to answer research questions and support the hypothesis. For instance, this study employs constructs such as task complexity, result control tightness, behavioral/action control tightness. Every contract is explained by several numbers of questions.
To the test the convergent and discriminant validity varimax principle rotated component analysis (hereafter PCA) was conducted on all constructs. In PCA model is built such that the scores on the questionnaire items (observed) depend on their expected underlying common factor (each of the five micro-attributes, respectively) as well as measurement error. The result of the PCA analysis is presented in Table 4.2. For the loading to be significant the valued should exceed 0.5 and only components with Eigenvalue more than one is selected. Additionally, the results of the validity test demonstrate that significant value for all elements is 0, and KMO which measures the sampling adequacy is 0.799, for satisfactory analysis it should exceed 0.5. The results of the PCA analysis with detailed description of each variable are provided in subsequent paragraphs of this section. Moreover, this section provides descriptive statistics of variables with indication of min, max and mean values.
Additionally, the Cronbach’s Alphas analysis which measures the reliability of the questions and to what extent the items that are supposed to measure single construct actually measures one and the same thing was conducted. The Cronbach’s Alphas of varibles are provided in the Table 4.1. And finally, this section provides the outcome of descriptive statistics where minimum, maximum and mean values of each construct are presented.
4.2.1 Task complexity construct
The construct for measuring task complexity is based on the work of Van den Ven and Ferry (1980). Questions are measured at the 5-point Likert-scale: strongly disagree, agree, neutral, agree, and strongly agree. There are overall seven questions designed to assess the task complexity. Four of which measure the task characteristics such as routines, interdependence, and completeness,
24 other three questions relate to the time and effort required to perform the task. This difference is reflected in the PCA, for instance first four questions 12-6, 12-5, 12-2 and 12-1, which are designed to measure task characteristics loaded on a single component, while other three questions 12-4, 12-3 and 12-7 which measure the effort and time needed to perform task loaded on another element. This study apply and assumption that that the task complexity can as due to the task characteristics as due to effort and time required to perform task, therefore in further analysis the all questions regarding task complexity will be included in the analysis. Additionally, questions 12-1, 12-2, 12-5, 12-6 are negatively worded and hence were reverse coded.
The Cronbach‘s Alpha for all items is 0.591 which is an acceptable number, however, to have more reliable measure it should be at least 0.7. According to the outcome of the analysis, Cronbach’s Alpha can be increased by deleting question 12-7. Thus question measures the amount of employees required to complete the task, which actually refers to the human capital intensity, therefore I decided to remove this question. After removing this question, the Cronbach’s Alpha is increased on 0.621.
4.2.2 Result control tightness construct
The construct designed to measure the result control tightness is composed of eight questions and measured at the 5-point Likert-scale: strongly disagree, agree, neutral, agree, and strongly agree. The questionnaire considers that control tightness can be increased by implementation of implicit and explicit controls, therefore tthe questions divided into two parts. For instance, the construct is measured by four questions that measure explicit tightness and the same number of questions that measure the implicit tightness. In the cornerstone of the questions regarding explicit tightness lays study of Van den Ven and Fery, (1980) and Bodewes (2000), implicit tightness, on the other hand, is based on the works of Van der Stede (2001) and Hage and Aiken (1968). These differences in approach are also visible in the factor analysis. Thus, the results of factor analysis show that all explicit result control and one implicit load on a single component, implicit controls load on separate single component and two other numeric control load on another single component. It should be noted that questions that were negatively worded have been reverse coded for the accuracy of the data analysis. It should be noted that questions 5-9, 5-10, 5-11 are negative worded and therefore are negatively worded.
The Cronbach‘s Alpha for all questions is 0.585, which is again acceptable level, but it should be at least 0.7 to have a reliable result. Higher alpha can be achieved by analyzing explicit and implicit control tightness separately. Thus, if only consider explicit control tightness, then the Cronbach‘s Alpha increases to 0.820 which represents the high quality of the question to measure
25 the construct. On the other hand, implicit control tightness is 0.233. Therefore, in further analysis only explicit result control is considered.
4.2.3 Action control tightness construct
As a basis to measure the concept of action control, the construct for measuring action control based on the assumption proposed by Hage and Aiken (1968), Van den Ven and Fery (1980), Bodewes (2000), Morgenson and Humphrey (2006) is developed. The construct consists of eight questions. When conducting PCA with eight items, five items loaded on the same component (one item with a negative number) and other the loaded on the same component as well. Explicit and implicit control tightness is the explanation for this difference. Thus, explicit control tightness measures the extent of use of standardized processes, procedures, rules and routines as part of the management control system, where a tight system is defined as one with a lot of controls regarding amount and scope. Implicit control tightness, on the other hand, The degree to which deviation from established processes procedures, rules, and routines is tolerated and/or encouraged, where a tight system is defined as one which does not allow any deviation from standard processes, procedures, rules and routines.
The Cronbach‘s Alpha for all questions is 0.539. However, if we explicitness and implicitness of the questions and consider each group of an item separately then The Cronbach‘s Alpha for explicit control tightness is 0.756, which indicates the that it is an appropriate instrument for measurement. On the other hand, implicit control tightness is 0.599. Therefore, based on this result for following analysis only explicit action control tightness will be considered.
4.2.4 Control variables
To obtain the unbiased estimate of the causal effect in the regression analysis and to prevent the omitted variable bias three types of control variables such as organization size, education level and gender are used of analysis.
4.2.4.1 Organization size
Among studies it is assumed that organization structure, especially organization size is contingent on the way how organization operates (Bruns and Waterhouse, 1975). Therefore, this variable is selected as a control variable to test the hypothesis. It is a newly adopted variable and it is measured by the total number of people employed by the organization. The survey consists of question where respondents had to provide information about a total number of employees based on four point scales: less than 100, more than 100 but less than 500, more than 500 but less than 5000 and more than 5000.
26 4.2.4.2 Education level
The respondents were asked to specify the highest level of education. Thus, employees had to range the education level as follows: Bachelor degree or lower, master degree and Ph.D. or other professional doctorate degree. In order to eliminate the misinterpretation the variable was grouped into two categories such as bachelor degree represents one category, master degree and PhD/higher represent second category.
4.2.4.3 Gender
One of the major demographic variables have been studies intensively in the previous literature is the gender of the respondents. Results of the studies of the gender difference by Buttner and Rose (1988) show that women are perceived having less leadership skills, less autonomy, less readiness to change. Another study by Meyers-Levy (1989) advocate that under separate circumstances women and men act differently in assessing information. So, based on these gender related differences, gender of the respondents is used as another demographic control variable in the analysis.
27
Constructs 1 2 3 4 5 6 7
Result control
Q5_6 Employee attainment of goals/targets is checked constantly. .813
Q5_7 My supervisor frequently checks to make sure that I am meeting my performance targets. .777
Q5_1 In my job, there is a performance measure for everything. .713
Q5_4 My organization sets a large number of performance goals/targets that I am expected to meet. .709
Q5_12 In my organization, employees are expected to meet pre-established goals/targets with no exceptions. .551
Q5_11 Responding to new, unforeseen opportunities is considered more important by my supervisor than achieving pre-established
goals/targets. .736
Q5_9 In our organization, goals/targets are essentially a guideline rather than a true commitment. .706
Q5_10 My supervisor is very considerate of my explanations of deviations from pre-established goals/targets. .468
Q6 Approximately how many performance targets are used in the evaluation of your job performance? .562
Q7 How often do you discuss the results of your performance measures with your supervisor? .531
Action control
Q4_11 The organization I work in primarily uses established processes, procedures and rules to give broad guidelines as to how activities are
to be performed. -.742
Q4_1 Whatever situation arises, we have existing processes, procedures or rules to follow in dealing with it. .731
Q4_2 Established processes, procedures and rules cover all of my job tasks. .720
Q4_4 In my organization, we have rules for everything. .707
Q4_8 My supervisor frequently monitors the extent to which I follow established process, procedures and rules. .517
Q4_12 Employees in my organization are encouraged to use procedures flexibly. .838
Q4_10 My job allows me to decide how to adjust rules to best perform my job tasks. .812
Q4_13 Employees in my organization are encouraged to adjust procedures to suit the situation. .810
Task complexity
Q12_6 I would describe my work as routine. .741
Q12_5 The situations, problems and issues that I encounter in performing my major tasks are usually the same. .732
Q12_2 Most of the time, I know what to outcome of my work efforts will be. .627
Q12_1 I can easily determine whether I have performed my work correctly. .503
Q12_4 In my work, I spend a lot of time solving difficult problems with no immediate solutions. .816
Q12_3 I often encounter problems in my work for which there are no immediate or apparent solutions. .771
Q12_7 My job depends on the work of many different people for its completion. .541
Components
28
5. Preliminary analysis
Notwithstanding to get a complete picture of the constructs descriptive statistics analysis was conducted. The results are presented in Table 5.1. The table provides minimum and maximum values as well as means of all variables that will be used in the hypothesis testing. It should be noted that all missing values have been removed before conducting the descriptive statistics, therefore, the sample size is decreased to 306 respondents.
Table 5.1 Descriptive Statistics
N Minimum Maximum Mean Deviation Std.
Employee age 306 19 63 34,69 7,883
Employee work experience 306 1 11 6,93 3,073
Organization size 306 1 4 2,84 1,084
Education level 306 1 3 1,74 642
Explicit behavioral control tightness 304 1 5 31,379 85,278
Explicit result control tightness 306 1 5 28,979 86,257
Task complexity 306 1 4,83 29,566 60,417
Valid N (listwise) 304
Furthermore, preliminary correlation was performed to test the relation between constructs. In conducting correlation testing Pearson and Spearman’s correlation techniques were used. This test is designed to provide the first indication of the bivariate relation between variables. The results of the analysis are provided in Table 5.2.
Since hypotheses consist of only one independent variable task complexity multicollinearity assumption which means close relation of several independent variables in the regression analysis is eliminated in this study.
30 1 2 3 4 5 6 7 -.072 -.191** .011 .259** .112 -.004 -.206 .001 .854 .000 .050 .938 -.076 .482** -.260** -.140* -.014 -.217** .188 .000 .000 .015 .806 .000 -.240** .497** -.159** -.129* .111 -.113* .000 .000 .006 .024 .052 .048 -.019 -.240** -.115* .073 .055 .655** .744 .000 .045 .201 .335 .000 .250** -.153** -.115* .090 .261** .016 .000 .007 .045 .118 .000 .786 .110 -.056 .116* .079 .261** -.022 .054 .331 .043 .169 .000 .705 -.035 -.214** -.124* .644** .037 -.028 .547 .000 .031 .000 .520 .626 Work experience
Pearson correlations appear below the diagonal, Spearsmans correlations appear above the diagonal *. Correlation is significant at the 0.05 level (2-tailed).
Task complexity
Explicit result control tightness
Explicit behavioral control tightness
Employee age
Education
Organization size
**. Correlation is significant at the 0.01 level (2-tailed).
31
6. Data analysis procedures and equations
Research hypotheses are tested by application of the ordinary least squares regression model. Since study aims to investigate the effect of the task complexity on MCS, the regression equation uses explicit result control tightness construct and explicit behavioral control tightness construct as a dependent variable, while task complexity construct as an independent variable. Additionally, to capture the moderating effect employee age and employee work experience in the current field are applied. Additionally, to better predict the effect of the dependent variable and decrease the effect of the error control variables such as organization size, education level and gender are implemented in the regression test.
Furthermore, to examine whether the relation between dependent, independent variables and moderator variables will differ under different condition additional variables such as organization size and level of education are included in the analysis. Thus, the following first two equations are designed to examine the effect of task complexity on the explicit result and behavioral control tightness. While the unstandardized values are used in the equation and additional parameter β0, which shows the value of the dependent variable when the value of the independent variable is zero, is added to the equation:
(1) Explicit behavioral control tightness = β0+ β1*TC + Age + Exp+ OS + EdL + Gen + ε (2) Explicit result control tightness = β0 + β1*TC +Age + Exp+ OS + EdL+ Gen + ε
Next step is to examine the effect of moderators on the interaction between task complexity and control dependent variables. Therefore, employee age and work experience are added to the regression equation. With an aim to have the more accurate result of testing standardized values of the task complexity, age and work experience are used in forming the interaction effect. Therefore, necessary z-values were calculated for these variables. The final regression models are as follows:
(3) Explicit behavioral control tightness = β0 + β1*TC + Age + Exp + β2*Age + β3*TC*Age + OS + EdL + Gen + ε
(4) Explicit result control tightness = β0 + β1*TC + Age + Exp + β2*Age + β3*TC*Age+ OS + EdL + Gen + ε
Meaning of the abbreviation used in the regression analysis and subsequent tables: TC stands for task complexity, EBCT – explicit behavioral control tightness, ERCT – explicit result control
