The effect of management control packages on management control
effectiveness and in turn on autonomous motivation
Master Thesis Accountancy
University of Groningen, Faculty of Economics and Business
January 20, 2020
Tim van Leeuwen
Studentnumber: 2744821
Ganzevoortsingel 30A
9711 AM Groningen
tel.: +316 18 63 93 70
e-mail: t.j.j.van.leeuwen@student.rug.nl
Word count: 10806
Supervisors university:
Prof. dr. ir. P.M.G. van Veen-Dirks
&
1
ABSTRACT
The purpose of this research paper is to investigate the influence of management control packages on management control effectiveness and in turn, on autonomous motivation. The current paper explores how an organization should implement its management control package in order to be perceived as effective. By using the levers of control framework of Simons (1995), the paper assesses the optimal distribution between management control elements. Additionally, this paper builds upon the Self-Determination Theory to link management control effectiveness to autonomous motivation. The study retrieved data of a conducted survey with 215 respondents of two public organizations. Its focus is on the perception of those subjected to management control, without a managerial function. The analyses do control for the effect of contextual variables, the three basic work needs and the characteristics of the respondents. Results reveal that nearly balanced management control packages have a significant positive influence on management control effectiveness. The analyses provide a first impression on the optimal use of the levers of control in a management control package. Furthermore, no significant relationship is found between management control effectiveness on autonomous motivation. A limitation of the current study is the absence of a generally accepted measure for management control effectiveness. Future research should more comprehensively examine how management control packages should be used to improve effectivity, considering its context. Future research on autonomous motivation should focus on what affects the three basic needs, as these are strongly related to autonomous motivation.
2
INTRODUCTION
Management control should fit into the organization in order to stimulate employees to satisfy the objectives of an organization (Speklé, 2001). The management control elements implemented in organizations by managers have an impact on whether the organization achieves its objectives effectively. Management control elements are the systems, practices, activities, values and regulations used to manage employee behavior (Bedford, Malmi & Sandelin, 2016). All selected elements together form a management control package (Malmi & Brown, 2008). Malmi and Brown (2008) emphasize the importance of investigating management control “as a package”, instead of examining management control elements in isolation. Management control should be investigated as a package because management control elements interrelate and are dependent on the use or absence of other management control elements (van der Klok, 2016). As prior research often neglected the effect of interrelation, the existing literature does not offer a representation of how management control operates within its organizational context. The present study will provide insights on how management control elements can be simultaneously used to complement each other and considers their interrelation in a management control package.
Management control elements can be classified in opposing groups, facilitating controls and constraining controls, to investigate the effect of interrelating forces (Simons, 1995). The opposing groups complement each other. When the two opposing groups have allocated a similar amount of management control elements, the management control package is in balance. A management control package needs to be well-balanced in order to direct employee behavior to satisfy the objectives of an organization (Mundy, 2010).
The effect of a management control package should be measured to examine whether the management control package has the desired effect. The most important function of management control is to align employees' behavior towards intended objectives (Speklé, 2001). Prior literature evaluated the achievements of management control packages through measures of firm performance (Langfield-Smith, 2008). However, not all organizational objectives are linked to firm performance and many factors can affect firm performance. Therefore Bedford, Malmi and Sandelin (2016) defined management control effectiveness to measure the direct results derived from a management control package. Management control effectiveness is achieved when a management control package can fulfill the three main functions of management control: goal alignment, adaptability and integration. The present study will examine whether a balanced management control package can influence the perceived management control effectiveness by employees. Investigating this relationship provides a relevant extension to prior literature. First of all, because the current study uses quantitative data to test the influence of a balanced management control package. In contrast to prior studies (Kominis and Dudau, 2012; Mundy, 2010), which use qualitative data to show the effect of a balanced management control package. Secondly, the present study identifies a way to measure whether a management control package has the desired effect through the use of management control effectiveness. There is no generally accepted
3
measure to capture the effect of a management control package. This paper suggests to use management control effectiveness to improve the understanding of management control packages, based on the definition of Bedford, Malmi and Sandelin (2016).
Prior literature (Fu, Hsieh & Wang, 2019) primarily focuses on the effect of management control on cooperative employee behavior, because organizations require employees to behave in favor of organizational objectives. The relation to employee motivation is, to a great extent, unexplored. When management control effectiveness is perceived as effective, the individual goals of employees align with the organizational objectives. Goal alignment could stimulate employee motivation because employees feel more identified with the objectives of the organization. The fulfillment of adaptability and integration, two of the main functions of management control effectiveness, can further develop the motivation of employees. The Self-Determination Theory, defined by Deci and Ryan (2008), enables us to examine how contingencies facilitate the enhancement of employee motivation in a work context. Prior research did address motivation as a unitary concept or exclusively defines intrinsic and extrinsic motivation (Lee, 2019). The Self-Determination Theory, used in the present study, refers to motivation as a continuum. A process called internalization occurs when external regulation transfers to intrinsic behavior. This process facilitates autonomous motivation (Deci & Ryan, 2008). The current study will investigate the effect on autonomous motivation of employees when a management control package is perceived as effective.
This study thereby further examines the perception of employees, as on an organizational level is often focused on employee behavior and the used management control elements. Management control literature generally examines the effect of management control based on responses from managers (Baerdemaeker & Bruggeman, 2015; Fu, Hsieh & Wang, 2019). However, Tessier and Otley (2012) argue that the perception of regular employees to management control could be different from the opinion of managers on how they use management control.
Prior literature did never use management control effectiveness to capture the effect of management control packages and in turn, to study its influence on autonomous motivation. Furthermore, insights will be provided on the balance of a management control package to indicate its effect on management control effectiveness. It leads to the following research question:
How do management control packages influence management control effectiveness and in turn, autonomous employee motivation?
The present thesis is structured as follows. The second section provides the theoretical background on management control packages, management control effectiveness and autonomous motivation to develop the hypotheses. Next, the third section describes the research methodology. In the fourth section, the findings are presented and analyzed. Finally, the fifth section comprises the discussion and presents the conclusions and recommendations.
4
LITERATURE REVIEW
Management control package
A management control package should stimulate employee behavior towards the organizational objectives (Speklé, 2001). The appropriate management control package, consisting of various management control elements, should be implemented to enable management control effectiveness (Merchant & van der Stede, 2007). There are various frameworks available to classify management control elements in groups (van der Kolk, 2016). Merchant and van der Stede (2007) use their object-of-control framework to distinguish various types of management control elements based on action controls, personnel control, culture control and results controls. Simons (1995) introduced the levers of control framework, which consists of four groups of management control elements to control an organization. Both frameworks seem to be focused on studying management control elements “as a package” (van der Kolk, 2016). However, the levers of control framework of Simons (1995) has more emphasis on the interrelation between various management control elements compared to the object-of-control framework (Mundy, 2010). The levers of control should be balanced in order to create dynamic tension and, therefore, to be effective (Widener, 2007). The balance between positive controls, facilitating management control elements and negative controls, constraining management control elements, is responsible for the dynamic tension (Mundy, 2010). It is a balance between keeping control over the organization's strategic goals and enabling autonomy to employees to let them seek opportunities (Sprinkle, 2003). Simons (1995) demonstrates that dynamic tension can enable an effective way to control organizational strategic objectives. The focus on interrelation between positive and negative controls makes this framework relevant, as the appropriate balance is needed in a particular organizational context. Mundy (2010) indicates that there will be an optimal distribution for the levers of control, in which the opposing groups complement each other the most. Furthermore, the levers of control specifically focus on activities to achieve strategic goals (Simons, 1995). Therefore the use of levers of control in an environment of complex objectives, like the public sector, could be relevant to a greater extent.
Management control elements can be classified into the levers of control, defined by Simons (1995). All the management control elements together form a management control package (Malmi & Brown, 2008). The implemented management control package should be well-balanced in order to accomplish the defined objectives (Merchant & van der Stede, 2007). The positive controls, the facilitating management control elements, consist of belief systems and interactive control systems. These are referred to as “positive and inspirational forces” (Simons, 1995). Positive controls enable the search for opportunities. Belief systems illustrate the internal core values reflected in the culture of the organization. Belief systems are depended on the intrinsic involvement of employees in an organization’s vision and mission and thus focusses on strategy. Interactive control systems represent the continuous dialogue between employees and management. The focus of interactive control systems is on short-term learning and innovation in
5
the organization. Hence, its emphasis is primarily on performance. The other two levers of control, the constraining management control elements, consist of boundary systems and diagnostic control systems. These are negative controls, as these tend to give a more forced and controlled feeling to employees. Boundary systems reflect formal regulation and guidelines of an organization, primarily focused on how employees should not behave. Diagnostic control systems represent the financial and non-financial measures of performance of employees and its direct transparent association with rewards or punishments. The target of diagnostic control systems is to increase performance (Simons, 1995).
Figure 1: Levers of control (Simons, 1995)
Sandelin (2008) states that the appropriate management control elements should be selected in a given context of an organization. Chenhall (2003) substantiates that the effectiveness of management control is dependent on its context and identifies specific contextual variables, existing of organizational environment, technology, structure, size, strategy and culture. The flexibility of organizational culture could, for example, determine the use of the levers of control, in particular, the adoption of belief systems (Heinicke, Guenther and Widener, 2016). The contextual variables developed from contingency-based research and are important to take into account when studying management control, as these variables affect the use and perception of management control (Chenhall, 2003).
For instance, a management control package in the public sector should, based on its context, satisfy the needs of a public organization to be successful. The paper of Kominis and Dudau (2012) encounters critics on traditional management control in the public sector. Traditional management control refers to a high association of diagnostic control systems. The high degree of diagnostic control systems does not fit into the needs of the public sector and should be complemented by more interactive control systems (Kominis & Dudau, 2012). Employees in the public sector seem to be attached to the public sector because of intrinsic reasons relatively more than employees in the private sector (van der Kolk, van Veen-Dirks & ter Bogt, 2018). Therefore the emphasis on extrinsic rewards offered by diagnostic control systems will not completely fit the needs of employees in the public sector. Furthermore, constraining controls can limit employees in their behavior and be perceived as a threat to their autonomy (Lee, 2019). When the facilitating and constraining controls are well-balanced, it will enhance effective control over organizational
6
objectives (Mundy, 2010). This example shows the effect a management control package with predominately constraining controls can have when this package does not fit into the organizational context.
Management control effectiveness
Prior research measured the performance of a management control package by use of financial indicators (Langfield-Smith, 2008). However, financial performance is not the singular objective of organizations and can be influenced by many factors. Bedford, Malmi and Sandelin (2016) defined management control effectiveness, which measures the performance of a management control package by rating the capacity of the management control package to fulfill its three main functions: goal alignment, adaptability and integration. Goal alignment indicates that management control should coordinate the objectives of individual employees and align their goals with the organizational objectives. Organizations should have a well profound view of their objectives. Adaptability refers to the flexibility of management control, whether it can quickly respond to external factors. Integration relates to the capacity of management control to cooperate between units to achieve goals (Bedford, Malmi & Sandelin, 2016). The accomplishment of these three functions together results in management control effectiveness.
Levers of control: Diagnostic control systems and boundary systems
An optimal balance between the levers of control in a management control package should create dynamic tensions to accomplish the highest management control effectiveness. The constraining negative controls, i.e. diagnostic control systems and boundary systems, are implemented to keep control over strategic goals by the use of extrinsic rewards or punishments (Sprinkle, 2003). Various direct activities could require extrinsic incentives to notice cooperative employee behavior (Yousaf, Yang & Sanders, 2015). Hofstede (1981) indicates that these activities primarily concern direct tasks with observable outputs. This kind of activities occur more often in the private sector, as in the public sector (Hofstede, 1981). Constraining controls should be used with caution, as these controls attempt to stimulate extrinsic motivation. Based on the crowding-out theory of Frey (1994), extrinsic motivation could harm existing intrinsic motivation. Furthermore, a high degree of constraining controls compared to facilitating controls can be perceived as a threat to autonomy for employees (Lee, 2019).
When a management control package is perceived as predominately negative, hence more constraining controls relatively to facilitating controls, it could negatively affect management control effectiveness. Goal alignment will be affected as employee behavior will be performed because of personal extrinsic rewards rather than their intrinsic reasoning and identification with the organizational objectives. It will cause the objectives of the organization and the goals of the employees to drift apart. Furthermore, a predominately negative management control package can limit employee behavior and, therefore, could negatively influence adaptability, another primary function of management control. The degree of use of constraining controls should fit the needs of
7
the context the organization operates in. Besides, constraining controls should be complemented with facilitating controls to not harm organizational objectives.
Levers of control: Interactive control systems and belief systems
To create a well-balanced management control package, negative controls should be complemented by positive controls (Mundy, 2010). The facilitating positive controls, i.e. interactive control systems and belief systems, explicitly highlight the profound values and beliefs of the organization, its culture and its activities. These positive controls focus on intrinsically motivated beliefs. The use of facilitating controls might have a positive effect on goal alignment. The values established in the culture of an organization should inspire employees to behave in favor of organizational objectives. It will enable the goals of organizations and employees to be more congruent. Furthermore, interactive control systems and belief systems provide opportunities to employees, in contrast to negative controls which primarily provide constraints (Sprinkle, 2003). Interactive control systems contribute to a dynamic environment of continuous learning in which employees can have autonomy (Kominis & Dudau, 2012). The created dynamic environment by positive controls could enable adaptability, a second primary function of management control. Furthermore, if objectives are more align between employees and the organization, the gap between goals of employees might be reduced. The goals of employees are more aligned with the objectives of the organization and therefore, the objectives of other employees. It could improve the coordination between units. Hence, the use of interactive control systems and belief systems can positively influence integration, the third function of management control.
However, when a management control package is predominately positive, which means more facilitating controls relatively to constraining controls, the control over the organizational objectives could be lost. The organization might lacks rules and monitoring to steer employee behavior. The degree of use of facilitating controls should, similar to constraining controls, match to the organizational context. Facilitating controls should be complemented with constraining controls to create an effective management control package.
Balanced management control package
According to Merchant and van der Stede (2007), the appropriate management control package should be implemented in order to accomplish management control effectiveness. An appropriate management control package means it should be well-balanced between facilitating and constraining controls (Mundy, 2010). In an optimal balanced management control package, the two opposing groups of management control elements should complement each other the most to reach the highest management control effectiveness. The balanced management control package should fulfill the main functions of management control and in turn, satisfy the needs of the organization in its context. Based on the described context on the levers of control, it leads to the following hypothesis:
H1 A balanced management control package has a positive effect on management control effectiveness.
8
Motivation
An important undiscovered aspect of management control is how employees are motivated. Individuals will act in a certain way, as they get motivated to choose a specific behavior (Vroom, 1964). Motivation is a crucial factor in achieving cooperative employee behavior. That means that motivation can ensure that organizational objectives are met (Zheng, Wang, Liu & Mingers, 2019).
Motivation can exist of intrinsic motivation, driven by internal personal values and extrinsic motivation, encouraged by external factors (Tadić Vujčić, Oerlemans & Bakker, 2017). The Self-Determination Theory of Deci and Ryan (2008) provides a broader distinction of motivation by their definition of autonomous motivation and controlled motivation. Deci and Ryan (2000) show that autonomous motivation and controlled motivation do not present a clear distinction as to how intrinsic and extrinsic motivation do. Motivation could be reflected as a continuum when the extrinsic motivation is more internalized the continuum shift towards autonomous motivation. Autonomous motivation consists of intrinsic motivation, integrated extrinsic motivation and identified extrinsic motivation. Controlled motivation consists of extrinsic motivation and introjected extrinsic motivation (Deci & Ryan, 2008). According to Gangé and Deci (2005), internalization is the process of transferring external values and regulation to internal value and behavior. Internalization appears in three stages in extrinsic motivation: introjection, identification and integration. When there is no internalization in extrinsic motivation, the behavior is only performed to satisfy external demand (Deci & Ryan, 2000). Introjection is the inferior stage in which employees take in external values, but these do not feel like their own. Their behavior is driven by contingent self-esteem (Deci & Ryan, 1995). Extrinsic motivation based on no internalization or introjection feels forced for employees, as behavior is not internally driven and therefore are associated with controlled motivation. The second stage of internalized extrinsic motivation is identified extrinsic motivation in which employees perceive the performed behavior as an aspect of their own beliefs; their behavior is aligned with personal goals (Gagné & Deci, 2005). The fullest stage of internalization of extrinsic motivation is integrated extrinsic motivation, employees fully identify themselves with the performed behavior and honestly believe their behavior is “instrumentally important for personal goals” (Gagné & Deci, 2005). Integrated extrinsic motivation is, however, still considered as extrinsic motivation as behavior is not performed because of personal satisfaction (Deci & Ryan, 2000).
9 Figure 2: Continuum of motivation (Gagné & Deci, 2005)
Management control effectiveness and motivation
Management control effectiveness means the implemented management control package is successful in fulfilling its three main functions (Bedford, Malmi & Sandelin, 2016). The current paper will examine whether the fulfillment of these functions increases autonomous motivation. The first main function is goal alignment. Goal alignment ensures the organizational objectives are congruent with the goals of employees (Bedford, Malmi, & Sandelin, 2016). Goal alignment could positively influence autonomous motivation when the defined organizational objectives give a feeling of identification or integration to the employees. If employees feel identified with the objectives of the organization, they can motivate their cooperative behavior based on intrinsic reasoning. If goals between the organization and the employee are not aligned, then employees require an external intervention, like rewards and regulation, to let them behave cooperatively. However, an external intervention can replace intrinsic motivation (Frey, 1994). So when extrinsic rewards are employed to accomplish cooperative behavior, the existing intrinsic motivation to carry out this behavior will diminish. Hence, the same level of external rewards must be employed to preserve the corresponding degree of cooperative behavior. In contrast to when goals are aligned between the organization and the employee, then cooperative employee behavior will be driven by intrinsic beliefs and the achievement of congruent goals might positively affect autonomous motivation. According to Greener (2019), management control should focus more on intrinsic motivation to preserve the intrinsic reasons the employees have for their cooperative behavior.
Another main function of management control is adaptability. Adaptability provides a feeling of an environment full of opportunities to employees. It could increase autonomy, which is one of the three basic work needs (Gagné & Deci, 2005). Gagné and Deci (2005) state that the three basic work needs, i.e. autonomy, competence and relatedness, are highly related to the process of internalization. Hence, adaptability might positively influence autonomous motivation.
The third main function of management control is integration. Integration refers to cooperation to achieve collective goals (Gagné & Deci, 2005). Integration could increase competence, as coordinated knowledge is used to accomplish objectives collectively. Competence is the second basic work need. Besides, integration could contribute to goal alignment between employees, as consensus on collective goals is required to implement integration. Harmony between employees could increase relatedness, the third basic work need.
The previously described context illustrates how the perceived fulfillment of the three main functions of management control might affect the autonomous motivation of employees. It leads to the following hypothesis:
10
RESEARCH METHODS
Survey
The present study will investigate the effect of management control packages, the management control elements are distinguished by use of the levers of control of Simons (1995), on management control effectiveness, determined by the fulfillment of the three main functions of management control described by Bedford, Malmi and Sandelin (2016), and in turn on autonomous motivation, defined by the Self-Determination Theory of Deci and Ryan (2008). The present study will make use of a database of a conducted survey to examine these relations.
The survey is conducted in 2017 among two universities of applied sciences and has 215 respondents. Hence, the survey is conducted in the public sector what could affect the results. Management control in the public sector has been questioned, because objectives of organizations in the private sector primarily focus on financial indicators, which are straightforward to measure, while objectives of organizations in the public sector are multi-dimensional focused on public value (Bao, Wang, Larsen & Morgan, 2013). Furthermore, the public sector more often includes indirect tasks that generate more complicated output to measure (Hofstede, 1981). The critics argued whether management control could take into account all the objectives and needs of the public sector, as these are more complex than in the private sector (Carter, Klein & Day, 1992). The distinction in objectives and activities of organizations between the public and private sector is a reason for a different approach to management control (Frey, Homberg & Osterloh, 2013). The public sector involves, because of its complexity, a highly relevant organizational context to examine how management control packages operate.
The 215 respondents are exclusively employees without managerial functions; these are educational staff and educational support staff. Managers have intentions to drive management control, while the current study is interested in the perception of those subjected to management control (Tessier & Otley, 2012). In this study, the perception of employees without managerial functions will be examined because management control packages are implemented to affect their behavior directly. The survey incorporated various questions based on prior literature to compose measures for variables used to analyze the relations.
Management control package
The management control elements are allocated to the four levers of control defined by Simons (1995). The levers of control themselves are divided by negative controls, the constraining controls, and positive controls, the facilitating controls. Ratios, the percentage of the score of a lever of control compared to the score of all levers of control together, are used to examine the levers of control relatively to each other. A ratio specifies the degree to which a lever of control is used compared to the total package. The two ratios of positive controls combined specifies the percentage of positive controls of the management control package and similar for the sum of the two ratios of negative controls.
11
Diagnostic control systems and interactive control systems are measured based on Bedford and Malmi (2015). Essential aspects of diagnostic control systems are to identify performance variables, set targets, monitor progress, adjust deviations and review critical areas of performance (Bedford & Malmi, 2015). Interactive control systems consist of continuous debates, the challenge of strategic uncertainties, involvement by dialogues (Bedford & Malmi, 2015). The elements of interactive control systems should be used by top management and operating management.
Boundary systems and belief systems are measured based on Kruis, Speklé and Widener (2016). Boundary systems depend on the code of business conduct. The organization should specify the appropriate behavior, behavior off-limits and risks that should be avoided in its code of business conduct. Additionally, the workforce should be aware of the code of business conduct (Kruis, Speklé and Widener, 2016). Elements of belief systems are the organization’s mission statement and core values. These should be communicated, so the workforce is aware of the mission statement and core values. Furthermore, the mission statement should inspire the workforce (Kruis, Speklé and Widener, 2016).
Management control effectiveness
Management control effectiveness is measured based on the fulfillment of the three main functions of management control, specified by the definition of Otley and Berry (1980). The three primary functions of management control are goal alignment, adaptability and integration (Bedford, Malmi & Sandelin, 2016). The current study examines the perceived management control effectiveness by employees subjected to management control (Tessier & Otley, 2012). The dimensions to which management controls should contribute in order to be effective, defined by Bedford, Malmi & Sandelin (2016), are (1) improving efficiency, (2) being innovative, (3) adapting to changes, (4) coordinating between units and (5) aligning subordinate actions. A 7-point Likert scale measures the accomplished contribution.
Autonomous motivation
Motivation is measured based on the continuum of the Self-Determination Theory defined by Gagné and Deci (2005). Gagné et al. (2014) use the Multidimensional Motivation at Work Scale to measure intrinsic motivation, identified extrinsic motivation, introjected extrinsic motivation and extrinsic motivation without internalization. Extrinsic motivation has been divided into material extrinsic motivation and social extrinsic motivation. The variable integrated extrinsic motivation has been left out of the study, as the initial scale could barely statistically distinct integrated extrinsic motivation from similar variables (Gagné et al., 2014). For each variable of motivation, three statements are used. Except for the variable introjected extrinsic motivation, which includes four statements. The degree of each variable of motivation is measured using a 7-point Linkert scale.
12
Control variables
In addition to the explanatory variables, the analyses include control variables to increase the reliability of the identified correlations. Control variables are used to check whether contributing factors can predict the variation of the dependent variable. The control variables consist of the general characteristics of the respondents of the survey. These include gender, age group, tenure group and educational background. The other control variables are based on prior literature.
Control variable: type of task
The current study identifies three types of tasks, namely direct tasks, indirect tasks and managerial tasks. These tasks are based on the functions of the respondents. The respondents do not include any managerial functions, as Tessier and Otley (2012) state the perception of those subjected to management control, regular employees, could be different from those who drive management control, managers. Hofstede (1981) emphasizes the difference between direct and indirect activities when investigating management control. Direct tasks are more observable and are therefore more applicable to management control. The educational staff will be referred to as employees with direct tasks and the educational support staff as employees with an indirect task. The variable type of task will be included in the analyses as a dummy variable.
Control variables: basic work needs
Three control variables based on prior literature consist of the basic work needs, defined by Deci and Ryan (2000). These are autonomy, competence and relatedness. The three basic work needs are essential to facilitate the process of internalization (Deci & Ryan, 2008). The three basic work needs are measured based on Van den Broeck, Vansteenkiste, De Witte, Soenens and Lens (2010). Autonomy refers to freedom on how to perform a task at work. Competence means that employees have the right knowledge to perform a task. Relatedness specifies the social connections of employees at work. All three of them are measured by the use of a 7-point Linkert scale.
Control variable: culture
Chenhall (2003) identifies culture as one of the key contextual variables to understand management control. All surveys of the current study have been conducted in the Netherlands, thus this study focuses on organizational culture instead of national culture. Two cultural variables are implemented as control variables in the analyses, namely group culture and developmental culture based on Kruis, Speklé and Widener (2016). Group culture concentrates on whether the organizational culture is referred to as a team environment and focuses on loyalty and morale. Developmental culture examines if the organizational culture is committed to innovation and ready to meet new challenges. These cultural variables are both a measure for flexibility of the culture, as defined by Heinicke, Guenther and Widener (2016). Flexibility expresses openness, adaptability and responsiveness of the organization, similar to group culture and developmental culture. A 7-point Linkert scale measures the variables.
13
Methodology hypothesis 1
The variables are all measured based on prior literature, as previously indicated, and will be used to perform the analyses. First will be evaluated whether the survey data is reliable to use in the analyses. Subsequently, three different regression analyses will be performed to examine the effect of management control packages on management control effectiveness and check whether hypothesis 1 can be accepted. Three statistical models are formed to analyze the effect of the control and explanatory variables on the dependent variable. First of all, the effect of each lever of control on management control effectiveness will be tested when implemented in a management control package (Malmi & Brown, 2008). A ratio is created of an individual score of a lever of control compared to the total score of the management control package. The output provides a view on the influence of the individual levers of control on management control effectiveness, taking into account the effect other levers of control have on management control effectiveness simultaneously.
Statistical model 1 hypothesis 1
MC effectiveness = β0 + β1 Belief to Package + β2 Interactive to Package + β3 Boundary to Package + β4 Diagnostic to Package +β5 Autonomy + β6 Competence + β7 Relatedness + β8 Group + β9 Development + β10 Task dummy + β11 Gender dummy + β12 Age + β13 Tenure + β14 Educational dummy + ε
Secondly, the current study will examine whether a balanced management control package has a positive effect on management control effectiveness. The second regression analysis uses the variable balanced package dummy to investigate the effect of a balanced management control package on management control effectiveness, in contrast to a predominate positive or negative focused package. A balanced management control package means that positive and negative controls both occupy nearly 50%. The model does take into account a margin based on the average deviation of 8.16% on both sides.
The variable is calculated by subtracting the ratio of total positive controls combined by the ratio of negative controls combined. When the outcome is zero or in between the average deviation, the management control package is in balance and the value of one will be given for the dummy variable. If the outcome for the management control package exceeds the average deviation, hence higher as 8.16%, the package is predominately positive focused. When the outcome exceeds -8.16%, the package is predominately negative focused. A predominately positive or negative focused management control package will be given the value zero for the dummy variable.
An example, a management control package has the following distribution of ratios for the levers of control: .35 belief systems, .20 interactive control systems, .25 boundary systems and .20 diagnostic control systems. This management control package has .35 (belief systems) + .20 (interactive control systems) = .55 of positive controls and .25 (boundary systems) + .20 (interactive control systems) = .45 of negative controls. The following formula will indicate whether this management control package is in balance or predominately positive or negative
14
focused: .55 - .45 = .10. The outcome for the management control package exceeds .0816 and, therefore, this management control package is predominately positive.
Furthermore, additional analyses will be performed to test for robustness. Similar constructed management control package dummies will be analyzed. Nevertheless, these include different margins. Hypothesis 1 will be first analyzed with the average deviation of 8.16%, as previously described. Additionally, robustness will be tested for this variable by including similar models with dummies for management control packages balanced with margins of 1%, 5% and 10%.
Statistical model 2 hypothesis 1
MC effectiveness = β0 + β1 Balanced Package dummy +β2 Autonomy + β3 Competence + β4 Relatedness + β5 Group + β6 Development + β7 Task dummy + β8 Gender dummy + β9 Age + β10 Tenure + β11 Educational dummy + ε
The third regression analysis will perform additional tests to provide first impressions whether a particular higher ratio of an individual lever of control compared to another lever of control will have a positive or negative effect on management control effectiveness. Furthermore, it will test if a management control package with a higher ratio of positive controls compared to its ratio of negative controls affects management control effectiveness. The analyses will explore the optimal distribution of the levers of control.
Statistical model 3 hypothesis 1
MC effectiveness = β0 + β1 Ratio of one lever of control compared to another lever of control +β2 Autonomy + β3 Competence + β4 Relatedness + β5 Group + β6 Development + β7 Task dummy + β8 Gender dummy + β9 Age + β10 Tenure + β11 Educational dummy + ε
Methodology hypothesis 2
For hypothesis 2, there will be evaluated first whether the collected data for autonomous motivation is reliable. Integrated extrinsic motivation has been dropped out, based on Gagné et al. (2014). The Cronbach’s alpha of autonomous motivation is calculated by encompassing the questions of intrinsic and identified extrinsic motivation. Subsequently, the regression analysis will be performed to investigate the effect of management control effectiveness on autonomous motivation and check whether hypothesis 2 can be accepted. The variable balanced package dummy will be included in the model as a control variable to test for a direct effect of balanced management control packages on autonomous motivation. Furthermore, the regression analysis for hypothesis 2 contains the same control variables as those used for hypothesis 1.
Statistical model 1 hypothesis 2
Autonomous motivation = β0 + β1 MC effectiveness + β2 Autonomy + β3 Competence + β4 Relatedness + β5 Group + β6 Development + β7 Task dummy + β8 Gender dummy + β9 Age + β10 Tenure + β11 Educational dummy + ε
15
Figure 4 Measurement of variables Dependent variables
Variable Description Measurement
MC effectiveness
Autonomous
Management control effectiveness
Autonomous motivation
The perceived fulfillment of the three main functions of management control: goal alignment, adaptability and integration (Bedford, Malmi & Sandelin, 2016).
Consists of intrinsic motivation, integrated extrinsic motivation and identified extrinsic motivation based on the continuum of the Self- Determination Theory by the use of the Motivation at Work Scale (Gagné et al., 2014).
Independent variable Explanatory variable
Variable Description Measurement
Balanced package dummy
Management control package in balance
Compares a nearly balanced package, 50% positive controls and 50% negative controls with taken into account an average deviation margin (8.16%), compared to a predominate positive or negative focused package.
Belief to pack
Boundary to pack Diagnostic to pack Interactive to pack
Ratio belief systems
Ratio boundary systems Ratio diagnostic control systems
Ratio interactive control systems
The inspirational core values of the organization measured based on Kruis, Speklé and Widener (2016).
The code of business coduct measured based on Kruis, Speklé and Widener (2016).
The process of controlling performance indicators measured based on Bedford and Malmi (2015). The process of employee involvement measured based on Bedford and Malmi (2015).
Control variables
Variable Description Measurement
Autonomy Competence Relatedness Autonomy Competence Relatedness
The three basic work needs based on Van den Broeck, Vansteenkiste, De Witte, Soenens and Lens (2010).
Group
Development
Group culture
Developmental culture
Level of group and developmental culture based on Kruis, Speklé and Widener (2016).
Task dummy Direct task
Compares educational staff with direct tasks to educational support staff with indirect tasks.
16
General control variables
Variable Description Measurement
Gender dummy Age
Male Age
Compares male to female respondents. The age of the respondents.
Tenure
Educational dummy
Organizational tenure Bachelor and Master degree
The tenure of the respondents in the organization. Compares respondents with a bachelor or master degree to respondents with a secondary education and secondary vocational education.
17
FINDINGS
In this section, the outcome of the research is described. First of all, the descriptive statistics of the 215 respondents employed at the universities of applied sciences. Figure 5 provides insights into their characteristics. The age of the respondents is divided over the age groups with an average of 48 years. Only a few of the respondents are under 30 years old. Furthermore, half of the respondent has worked at the organization for longer than ten years. Most respondents have a high-level education, i.e. bachelor or master degree. Besides, 65% of the respondents are educational staff and the other 35% are educational support staff.
Figure 5 Descriptive statistics N = 215
Gender Frequency Percentage
Male 102 47%
Female 113 53%
Age groups Frequency Percentage
20-29 years 6 3%
30-39 years 48 22%
40-49 years 56 26%
50-59 years 68 32%
60 + years 37 17%
Tenure groups organization Frequency Percentage
0-4 years 59 27% 5-9 years 50 23% 10-19 years 63 29% 20-29 years 30 14% 30-39 years 11 5% 40 + years 2 1%
Tenure groups department Frequency Percentage
0-4 years 78 36% 5-9 years 58 27% 10-19 years 54 25% 20-29 years 18 8% 30-39 years 6 3% 40 + years 1 0%
Education Frequency Percentage
Primary education 0 0%
Bachelor degree 61 28%
Master degree or higher 129 60%
Secondary education 4 2%
Secondary vocational education 21 10%
Function Frequency Percentage
Educational staff 139 65%
18
The survey was initially distributed on 17-10-2017 at 8 o’clock at the first university of applied sciences and ended on 27-11-2017. Afterward, the survey was distributed on 28-11-2017 at the second university of applied sciences and the last survey here was conducted on 13-12-2017. The survey of the first organization captured data of 140 respondents and the second organization had 75 respondents. Analyses are performed to check whether the retrieved data includes non-response bias or significant differences between the two organizations. An analysis of the early respondents versus the late respondents is performed to check for non-response bias, for both organizations separately. 50% of the late respondents can be considered as a proxy for non-response.
Figure 6 illustrates the difference in mean of the variables. The means of the first and second organization are shown in bold. Subsequently, the means of the early and respondents are presented separately for both organizations. An additional independent t-test is performed to compare the means of the respondents of the first and second company and the means of early and late respondents. No significant differences are found between the two organizations. However, some conspicuous significant differences stand out between early and late respondents of the second organization. The variables boundary systems (t (71.528) = 1.708, p = .091), interactive control systems (t (72.953) = 1.896, p = .062) and management control effectiveness (t (2.401) =2.401, p = .019) have significant differences in its’ means. Therefore some variables of the second organization include a concern for non-response bias. If the whole sample would have responded to the survey, there could be different results. Furthermore, the second organization does contain a highly significant difference in mean for the variable autonomous motivation, as t (134.275) = -5.071 and p = .000. The second organization thereby has a concern of non-response bias for the second hypothesis in particular. The existing non-response bias should be considered when interpreting the results and might indicate that outcomes of the research are more complicated to validate and generalize.
The means of the two organizations provide some first impressions on the balance of management control packages. The ratio of positive controls of the first organization is .482 calculated by (3.66 + 3.35) / 14.54 and a ratio of .518 allocated to negative controls calculated by (3.80 + 3.73) / 14.54. The management control package of the first organization is in balance since .482 - .518 = -.036, which does not exceed the average margin of .0816. The second organization represents similar results as the ratio of positive and negative controls is .489 and .511, respectively. With an outcome of -.022 the management control package of the second organization is, on average, in balance.
19
Furthermore, the descriptive statistics of the variables are examined to check for outliers and the distribution of the variables. There was only one case of an outlier for the variable belief systems, as one respondent gave all the full points at a 7-point Linkert scale for this variable. However, the data of the respondent is visually inspected and there is no suspicious reason to exclude or winsorize this data. In addition, the value of skewness and kurtosis are examined to check whether the variables are normally distributed. The maximum value for skewness, -.344, is acceptable as it is between -2 and 2. Therefore the variables are normally distributed and not asymmetrical. Besides, the maximum value for kurtosis is -.862, which means the variables do not have a sharp central peak but are more evenly distributed. The kurtosis value is acceptable, as it is between -7 and 7.
Additionally, the variables which consist of multiple 7-point Linkert questions are examined to assess the reliability of the variables with the use of its’ Cronbach’s alpha. The overall Cronbach’s alpha of autonomous motivation is calculated by encompassing the questions for both, intrinsic and identified motivation. As shown in figure 7, all variables have a Cronbach's alpha higher than 0.7. Hence, the internal consistency of all questions is valid.
Figure 7 Variable reliability
Variable
Cronbach's
alpha Control variable
Cronbach's alpha MC package Autonomy .882 Belief .871 Competence .804 Boundary .846 Relatedness .857 Diagnostic .965
Interactive .897 Culture group .815
Culture developmental .875 MC effectiveness .891 Autonomous motivation .832 Intrinsic .890 Identified .813
Figure 6 Means of first organization versus second organization and early respondents versus late respondents N = 215 Variable First organization (N=140) First early respondents (N=70) First late respondents (N=70) Second organization (N=75) Second early respondents (N=38) Second late respondents (N=37) Belief 3.66 3.55 3.78 3.77 3.93 3.60 Boundary 3.80 3.80 3.79 3.72 4.03 3.39 Diagnostic 3.73 3.76 3.70 3.85 3.92 3.78 Interactive 3.35 3.38 3.33 3.46 3.72 3.19 MC effectiveness 3.65 3.60 3.70 3.93 4.28 3.56 Autonomous 5.55 5.20 5.91 5.44 5.50 5.39
20
The matrix in figure 8 shows a Pearson correlation. The matrix is used to check whether various variables correlate reciprocally and to provide a first impression of the correlation between explanatory and dependent variables. The mutual correlation between interactive and diagnostic is .77, which is higher than .70 and, thus, could involve bias. Therefore, in particular, the VIF value of these variables should be considered when performing a regression. The other explanatory variables appear to correlate less than .70. Thus, it can be assumed that multicollinearity is not a concern for the other variables when the VIF values do not exceed 10. Furthermore, the first impression is that the four levers of control do significantly correlate individually with management control effectiveness. In the regression analysis, the levers of control will be further investigated as a package. Besides, the correlation between management control effectiveness and autonomous motivation does appear to be non-significant and low.
The regression analyses for hypothesis 1 are presented in appendix A. Figure 9 shows the regression analysis of statistical model 1 for the effect of a ratio of an individual lever of control on management control effectiveness. The model investigates the effect of the individual levers of control “as a package” since all levers of control are analyzed in one regression analysis and the variables reflect the levers of control relatively to each other, which makes them dependent on the presence of other levers of control. Figure 10 shows the regression analysis on the dependent variable management control effectiveness affected by the balanced management control package dummy. In addition, dummies with various margins are tested for robustness. In figure 11, several analyses are performed to check whether a particular higher ratio of an individual lever of control compared to another lever of control will have a positive or negative effect on management control effectiveness.
All figures are tested using the least-squares method in linear regression. The regression analysis of figure 9 includes two models, the regression analysis of figure 10 includes six models
Figure 8 Pearson correlation matrix
Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) Belief (2) Boundary .361** (3) Diagnostic .364** .438** (4) Interactive .550** .489** .770** (5) MC effectiveness .482** .446** .385** .544** (6) Autonomous .210** -.009 .041 .074 .006 (7) Autonomy .460** .254** .215** .416** .248** .368** (8) Competence .089 .048 .122 .107 .065 .325** .270** (9) Relatedness .307** .216** .135* .207** .219** .301** .434** .311** (10) Group culture .468** .261** .211** .368** .337** .237** .524** .165* .539** (11) Developmental .459** .239** .242** .394** .405** .268** .493** .143* .442** .622** (12) Task dummy -.183** -.206** -.100 -155* -.205** .122 -.089 -.081 -.083 -.109 -.108
** and * indicate that coefficients are statistically significant at the 1% and 5% level, respectively. Significance is based on two-sided testing.
21
and for figure 11 five models are included. For all figures, an analysis is first performed on the control variables only (figure 9 model 1, figure 10 model 1 and figure 11 model 1). Conspicuous is the significant effect of group culture, developmental culture and gender in the control variables only models (β = .163, p = .068 ; β = .298, p = .000 ; β = -.192, p = .003 ; respectively). It indicates that these are important variables to consider when examining the effect on management control effectiveness. The negative effect of male respondents on management control effectiveness could create bias if the survey is primarily conducted by one gender. In the present study, the number of male respondents (47%) and female respondents (53%) is nearly equal. Furthermore, Chenhall (2003) does emphasize that management control is dependent on its context and specifies culture as an important contextual variable. Culture should be taken into account when researching management control, as otherwise models could include bias. Subsequently, after the control only model, explanatory variables are added into the models.
Statistical model 1 hypothesis 1 (Appendix A figure 9)
Figure 9 model 2 concerns the full model of statistical model 1 and includes all the levers of control. The adjusted coefficient of determination, adjusted R-squared, in the full model of figure 9 is .284. The adjusted R-squared indicates how much of the variation of the dependent variable management control effectiveness can be predicted by the explanatory variables implemented in the model. The F-value indicates whether the explanatory variables reliably predict the dependent variable. When the F-value is close to zero, there is a higher chance that the null hypothesis is true and the explanatory variables have no effect on the dependent variable. The F-value of the full model of figures 9 is 7.051 and is significant as p = .000. Hence, the explanatory variables do affect the dependent variable. Furthermore, the highest variation inflation factor (VIF) is considered to rule out multicollinearity. Previous impressions by the Pearson correlation in figure 8 found a high mutual correlation between interactive control systems and diagnostic control systems. However, the VIF values for these variables are very low (VIF < 3). The variable boundary systems has the highest VIF in figure 9 (VIF = 2.955). The highest VIF does not exceed ten. Therefore, there is no concern for multicollinearity.
Figure 9 model 2 shows that belief systems, interactive control systems and boundary systems all have a significant standardized relationship with management control effectiveness (β = .202, p = .043 ; β = 0.342, p = .000 ; β = .399, p = .000 ; respectively). The relative use of these levers of control have a positive effect on management control effectiveness. The variable diagnostic control systems has no significant correlation with management control effectiveness (β = .138, p = .134). The model indicates which levers of control have the most effect on management control effectiveness, when implemented in a package.
Statistical model 2 hypothesis 2 (Appendix A figure 10)
Another regression analysis is shown in figure 10 to examine the balance of management control packages by the use of six models. The first model is with control variables only, equal to figure 9 model 1. Figure 10 model 2 shows statistical model 2 of hypothesis 1. The model uses the
22
balanced package dummy to analyze the effect of a balanced management control package, which has a distribution of 50% positive controls and 50% negative controls with a margin of 8.16% based on the average deviation, compared to a predominate positive or negative focused package. 137 of the 215 respondents perceived the management control package implemented in their organization as balanced, based on the previously described definition. Model 2 shows a balanced management control package has a significant positive influence on management control effectiveness (β = .290, p = .000). These findings align with the study of Mundy (2010), which emphasizes the importance of balanced use of management control systems. The adjusted R-square is .278, which means that 27.8% of the variation of management control effectiveness can be predicted through the explanatory variables of model 2. The F-value is 8.498 and significant (p = .000), which means the explanatory variables have an effect on management control effectiveness. Furthermore, multicollinearity is not a concern as the highest VIF is 2.140.
Furthermore, various additional analyses are performed to test for robustness. Similar variables are used as the balanced package dummy although the variables of models 3, 4, 5 and 6 consist of 2%, 5%, 10% and 15% margin, respectively. The management control package with a 10% margin shown in figure 10 model 5 has the highest positive significant coefficient (β = .347, p = .000). A management control package with a margin of 10% refers to a management control package with a ratio of positive controls and negative controls between .45 and .55. Additionally, model 5 has the highest R-square of .309 and the highest F-value with 9.685. When management control package have a broader or more narrow margin, the effect on management control effectiveness diminishes. What suggests that management control packages do not need to be precisely balanced in order to be most effective, nevertheless still management control packages should be nearly in balance.
Statistical model 3 hypothesis 1 (Appendix A figure 11)
Figure 11 shows the last regression analysis for hypothesis 1. Additional tests are performed to explore whether particular levers of control are preferred when these are compared relative to each other. The second model represents the influence of a management control package which has relatively more positive controls than negative controls. Model 2 shows a non-significant relationship for a management control package with relatively more positive controls (β = -.067, p = .308). That means a higher degree of positive control compared to negative control is not favored to improve management control effectiveness and neither the other way around.
Figure 11 model 3 up and till 5 include tests to examine whether a relatively higher or lower ratio between two levers of control could affect management control effectiveness. In model 3 a management control package with relatively more interactive control systems compared to diagnostic control systems has a significant positive effect on management control effectiveness (β = .142, p = .030). These findings are in accordance with the study of Kominis and Dudau (2012), as they state that diagnostic control systems should be complemented with more interactive control systems in the public sector. Furthermore, model 3 shows that a management control package with relatively more belief systems in contrast to boundary systems has a significant negative influence
23
on management control effectiveness (β = -.161, p = .012). This effect cannot be substantiated by prior literature.
Figure 10 model 4 examines for a cross-dimensional influence because belief systems and diagnostic control systems, as well as interactive control systems and boundary systems, have never been linked to each other by Simons (1995). Model 4 finds no significant relationships.
Figure 10 model 5 investigates the influence of a predominant distribution between the positive controls and between the negative controls. For the positive controls, a management control package with relatively more interactive control systems than belief systems has a positive relationship with management control effectiveness (β = .191, p = .003). For the negative controls, a positive effect of a management control package including relatively more boundary systems as diagnostic control systems is found (β = .134, p = .039).
The most revealing models of figure 10 are model 3 and 5, as these show significant relationships. These models have an adjusted R-squared of .233 and .237, respectively. The F-values are 6.828 and 6.529, respectively, and are significant as p = .000. The highest VIF of model 3 and 5 is 2.156, thus no concern for multicollinearity. These models indicate the optimal distribution for the levers of control in order to be perceived as most effective.
Statistical model 1 hypothesis 2 (Appendix B figure 12)
Figure 12 in appendix B shows statistical model 1 for hypothesis 2. The least-square method was conducted in a linear regression model. First, analysis is carried out on the control variables only, figure 12 model 1. The most outstanding are, as expected based on prior literature of Deci and Ryan (2000), the influence of autonomy, competence and relatedness on autonomous motivation (β = .272, p = .000 ; β = .235, p = .000 ; β = .131, p = .088 ; respectively). The findings confirm the Self-Determination Theory. Besides the three basic work needs, the dependent variable is also less significantly influenced by age and tenure (β = .159, p = .049 ; β = -.157, p = .045 ; respectively). A revealing outcome as it means that an older respondent has more autonomous motivation, nevertheless a respondent with a longer tenure has less autonomous motivation. The model does also test for a direct effect of a balanced management control package on autonomous motivation and found a non-significant relationship (β = -.104, p = .103).
Figure 12 model 2 examines the effect of management control effectiveness on autonomous motivation. The model shows a surprising non-significant negative influence of management control effectiveness on autonomous motivation (β = -.087, p = .232). The model has an adjusted R-squared of .233, which means the explanatory variables in the model can predict 23.3 % of the variable autonomous motivation. The adjusted R-square does, however, only rise .001 when management control effectiveness is added as a variable. Furthermore, the F-value drops from 6.863, in the control variables only model (figure 12 model 1), to 6.424 for the model which includes management control effectiveness. The highest VIF is 2.168. Thus, there is no concern for multicollinearity.
24
DISCUSSION AND CONCLUSIONS
In the current section, the findings described in the previous section will be evaluated against the hypotheses and the research question. Furthermore, this section will mention the limitations of the present study and provide recommendations for future research.
Hypothesis 1 states: A balanced management control package has a positive effect on management
control effectiveness.
The appropriate management control elements should be selected in a given context of an organization to create a well-balanced management control package (Merchant & van der Stede, 2007). A balanced management control package means 50% of the management control elements should be positive controls and 50% of the management control elements should be negative controls. The variable balanced package dummy used to capture the effect on management control effectiveness has a margin of 8.16%, equal to the average deviation. Figure 10 model 2 confirms hypothesis 1, as the balanced management control package dummy has a significant positive influence on management control effectiveness. Simons (1995) emphasizes the importance of complementing constraining forces of management control with facilitating forces of management control to create dynamic tension, that would allow adequate control of the organizational strategy. Based on the previously described findings, there could be statistically concluded that positive controls and negative controls in a management control package should be nearly equal in balance. A balanced management control keeps control over organizational strategic objectives and, besides, enables employees to seek opportunities (Sprinkle, 2003). Therefore, a balanced use of management control elements improves management control effectiveness (Mundy, 2010).
Additional analyses were performed in the current study to explore the optimal degree of the levers of control compared to each other. Figure 11 model 5 provides some first insights on which positive and negative controls should be selected to establish an effective management control package. A higher ratio of boundary systems in contrast to diagnostic control systems has a significant positive effect on management control effect, as well as a higher ratio of interactive control systems compared to belief systems. These findings contribute to a first impression on how management control elements should be selected to create an optimally distributed management control package. Nor only for the degree of positive controls compared to the negative controls, but to indicate the preferred use of all the levers of control. As these findings are not substantiated by prior literature, future research should explore these relationships to a greater extent.
Figure 11 model 3 further examines the optimal distribution of the levers of control. Similar to Kominis and Dudau (2012), the current study finds that a higher ratio of interactive control systems compared to diagnostic control systems has a significant positive effect on management control effectiveness. Kominis and Dudau (2012) found that diagnostic control systems in the public sector were overused and should be complemented by interactive control systems to be more effective. Furthermore, this study finds a significant negative effect between a higher ratio
25
of belief systems in contrast to boundary systems. Relatively more boundary systems, compared to belief systems, could improve management control effectiveness. This effect can be explained since belief systems primarily focus on facilitating a flexible culture, which can enable effectiveness (Heinicke, Guenther & Widener, 2016). Hence, by an indirect effect. Whereas boundary systems directly keeps control over organizational strategic objectives to improve effectiveness (Sprinkle, 2003). The distribution and reinforcing effect between belief systems and boundary systems should be further investigated by future research.
A developmental culture proved to be an important contextual variable for management control effectiveness, as it has a highly significant influence. The effect of culture as a contextual variable is in accordance with Heinicke, Guenther and Widener (2016). A limitation of the current study was that only a few contextual variables could be implemented in the analyses, because various relevant contextual variables, defined by Chenhall (2003) like structure and other cultural factors, were not reliable as a result of their low Cronbach's alpha.
Future research should implement additional contextual variables. A management control package should have the appropriate balance given its context, as Merchant and van der Stede (2007) state. Therefore, future analyses should include all contextual variables mentioned by Chenhall (2003). Additionally, future research could more comprehensively examine the optimal distribution of the levers of control to improve the perceived management control effectiveness. There are opportunities to explore this gap in the literature further using both, quantitative and qualitative data.
Hypothesis 2 states: Management control effectiveness has a positive effect on autonomous
motivation.
Management control is perceived as effective by employees when the three main functions, i.e. goal alignment, adaptability and integration, of management control are fulfilled, as described by Bedford, Malmi and Sandelin (2016). If management control is perceived as effective, it might positively influence autonomous motivation because the goal of employees would be more aligned with the objectives of the organization. As goals are more congruent, it would encourage the feeling of identification with performed behavior and stimulate the process of internalization (Gagné & Deci, 2008). The feeling of identification and integration would in turn, positively affect autonomous motivation. Therefore an effective management control package would have a positive influence on autonomous motivation.
The findings of the regression analysis in figure 11 model 2 show, in contrast to hypothesis 2, a surprising non-significant negative correlation between management control effectiveness and autonomous motivation. Based on the results, there could be concluded that management control perceived as effective by employees does not stimulate the process of internalization for these employees to attain autonomous motivation. The negative effect may occur due to a high perceived presence of management control when employees classify management control as effective. The recent study of Lee (2019) found that management control occasionally could be perceived as a
