Thesis working document
The impact of input controls, integrativity of services and innovation on autonomy in professional service firms.
Master Thesis
Name: Mark Witteveen
Student number: 5980917
Thesis supervisor: Dr. Ir. S.P. van Triest Date: January 28th, 2018
Program: MSc Accountancy & Control, specialization Control Faculty: Economics and Business, University of Amsterdam Word count: 13382, 0
Statement of Originality
This document is written by student Mark Witteveen 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.
Abstract
The purpose of this paper is to investigate a specific design feature of management control systems in professional service firms. More specific, this paper examines the effects of three determinants of autonomy: input controls, integrativity of services and innovation and the moderating effect of task uncertainty. Data analyzed in this study was obtained from surveys conducted with experienced managers working on the high and mid-management level at
professional service firms. The results provide partial evidence for the direct effect of innovation on autonomy. However, they also show that task uncertainty has no significant impact on two of these relations.
Keywords
Professional Service Firms, Management Control System; Input Controls, Integrativity of Services, Task Uncertainty; Autonomy
Contents
1 Introduction ... 6
2 Theory ... 8
2.1 Autonomy ... 8
2.1.1 Task based autonomy ... 9
2.1.2 Job crafting... 10
2.1.3 Alternative work arrangements and Idiosyncratic deals ... 11
2.1.4 Inhibiting factors for autonomy ... 11
2.2 Agency theory and delegation of decision rights ... 12
2.3 Self-efficacy and self-determination theory ... 13
2.4 Input controls ... 14
2.5 Integrativity of Services ... 15
2.6 Innovation ... 16
2.7 Task uncertainty ... 16
2.7.1 Origin of Task Uncertainty... 16
2.7.2 Task Uncertainty due to: analyzability and many exceptions ... 17
2.7.3 Task Uncertainty as a result of contracting difficulties in PSF’s ... 18
3 Hypothesis development ... 19
3.1 The relation of Input Controls with Autonomy ... 19
3.2 The relation of Integrativity of Services with Autonomy... 19
3.3 The relation of Innovation with Autonomy ... 20
4 Research method... 23
4.1 Survey design ... 23
4.2 Survey variables ... 24
4.2.2 Independent variables ... 25 4.2.3 Interaction variable ... 28 4.2.4 Control variables ... 29 4.3 Statistical model ... 29 4.3.1 Equations ... 32 5 Results... 33 5.1 Control variables ... 34
5.2 Main model: test of Input Controls, Integrativity of Services, Innovation and Autonomy ... 34
5.3 Interaction effect of Task Uncertainty ... 34
5.4 Robustness test ... 35
6 Conclusion ... 37
6.1 Discussion of findings ... 37
6.2 Limitations... 38
6.3 Implications for future research ... 39
7 References ... 40
1 Introduction
Management control system (MCS) research has focused primarily on the manufacturing industry, even though the service industry has outgrown the traditional industries in modern economies. Further investigation has been called for by various scholars. In response to this lack of research, earlier studies chose a contingency based perspective (Chenhall, 2003). Auzair & Langfield-Smith (2005) and Ittner (2003) researched the effect of strategy on the development of MCS in professional service firms, while Sharma (2002) finds that the design of MCS is influenced by environmental uncertainty, size and budget system. MCS are comprised of four control dimensions: (1) behavior controls, (2) result controls, (3) personnel controls and (4) cultural controls (Ouchi, 1979; Snell, 1992). This paper focuses on elements of the MCS design of professional service firms and their impact on autonomy of employees.
The last decades have shown a fundamental change in the way we work, caused by the possibilities created by communication and information technology and leading to a shift from manufacturing to knowledge work and services (Chen & McDonald, 2015; Grant & Parker, 2009). This resulted in a step-by-step transfer of control from organizations to employees, as employees claimed and received more autonomy and decision power (Langfred C. W., 2004). As acknowledged by Langfred & Rockman (2016) this transfer of control can be much more visible in some organizations than others and depends on the industry, work, location and culture. It is to be expected that organizations that specialize in knowledge work in the setting of contemporary information and communication technology and use the resources of highly qualified staff, such as professional service firms (PSF’s hereafter), will be more compelled to extend autonomy, than organizations that are active in production or more simple service concepts. For this reason, it is important to understand which factors determine the extent to which autonomy is granted. This is especially relevant, “as the inherent purpose of an organization remains to provide control and stability” (Langfred & Rockman, 2016), so factors that reduce autonomy are present in every organization. Therefore, the research question I will answer in this paper is:
What are the determinants of employee autonomy in professional service firms?
I will investigate three relations that affect autonomy because of their specific relevance for PSF’s. First, I look at the relation between input controls and autonomy. Almost all PSF’s find themselves in a situation where they have to rely on input controls, due to the nature of the
work. So to ascertain how different levels of input controls affect autonomy, the following research question is proposed:
RQ1: “What is the impact of input controls on autonomy in professional service firms” Second, I study the Integrativity of Services and see how this relates to autonomy. Integrativity of Services is the amount of client interaction during service delivery and is especially characteristic for PSF’s, since every company or business line has made deliberate choices with regard to this interaction with the client in order to achieve the best performance. Exploring the impact of Integrativity of Services to autonomy, the following research question is proposed:
RQ2: “What is the impact of Integrativity of Services on autonomy in professional service firms”
Third, I will study Innovation, since PSF’s are expected to bring specific, specialized and up-to-date expertise and incorporate this into their advice and services. This is vital for their business model and needs to be improved continuously. In this study, Innovation is defined as the aspiration to be innovative or to promote an innovative approach. This leads to the following research question:
RQ3: “What is the impact of Innovation on autonomy in professional service firms” Earlier studies have presented explanations for determinants of autonomy, but no evidence based research has been identified. Using a survey with respondents from PSF’s I will present evidence of this relation, thus generating a contribution to existing knowledge. This paper is structured as follows: Chapter 2 gives an overview of existing theories on Autonomy, Input controls, Integrativity of Services and Task Uncertainty. Based on these theories, the hypotheses are developed in Chapter 3. Chapter 4 describes the research method and in Chapter 5 the results are presented. The conclusion and discussion can be found in Chapter 6.
2 Theory
Several theories from both economical and psychological fields illustrate the relations between Input Controls, Integrativity of Services and Innovation on Autonomy. Therefore, this section provides a brief introduction of agency theory, theory on delegation of decision rights, and the social cognitive or self-efficacy theory and then elaborates further on Autonomy, Input Controls, Integrativity of Services, Innovation and Task Uncertainty.
2.1 Autonomy
Autonomy used to be defined as follows: the degree to which the job provides substantial freedom, independence, and discretion to the individual in scheduling the work and in determining the procedures to be used in carrying it out (Hackman & Oldham, 1976). Nowadays, these characteristics about work and autonomy are no longer valid, as current forms of work have changed drastically as a result of rapidly evolving technology (Grant & Parker, 2009). The definition for individual autonomy still holds, but the area over which the employee has decision rights and freedom has expanded in different ways.
Autonomy fosters responsibility, making employees more result-oriented and focused on getting the job done properly. It is a job aspect which will prompt employee feelings of personal responsibility for the work outcomes. To the extent an employee has high autonomy in his job, the outcomes will depend increasingly on the individual's own efforts, initiatives, and decisions rather than on the adequacy of instructions from the boss or on abiding to a manual of job procedures. In such circumstances, the individual will feel strong personal responsibility for the success and failures that occur in the job.
To explore how work and autonomy have changed over time, and to understand the challenges faced by managers in organizations, I will outline some of these changes and explore what individual autonomy is today. Firstly, autonomy plays a bigger role in retention of talented employees. Workers do not show the same loyalty towards their employers as they used to, so retention of high performers who exceed client expectations is an important issue for PSF’s, especially and generally, autonomy is expected by workers and offered in order to be found attractive in the labor market (Langfred C. W., 2004). Autonomy is also linked to well-being, (Wu, 2015), motivation, satisfaction and engagement (Dwyer, Schwartz, & Fox, 1992; Vera, 2016) and lower turnover (Annink, 2012), characteristics wanted by employers, who consequently offer autonomy to stay competitive. Parker (2006) found that autonomy was
associated with greater role-breadth self-efficacy, which in turn linked to the proactive behaviors of idea implementation and problem-solving.
A second aspect is motivation to excel in the job and be proactive. Theories of team self-management suggest that autonomy motivates teams to make independent decisions that serve the interests of their tasks in the best possible manner, by giving them a greater sense of responsibility and accountability for their work (Haas, 2010). In contemporary working environments autonomy can take on many shapes and this reflects how ideas about work have changed in the past decades, having resulted in a step-by-step transfer of control from the organization to its employees. “This ceding of control—combined with increased uncertainty and ambiguity in many industries and markets—necessarily causes tension and stress for managers in organizations, as the inherent purpose of an organization remains to provide control and stability” (Langfred & Rockman, 2016). This transfer of control can be much more visible in some organizations than others and depends on the industry, work, location and culture. It is to be expected that organizations like PSF’s that specialize in knowledge work in the setting of contemporary information and communication technology and use the resources of highly qualified staff, professional service firms, will be compelled to extend autonomy to a greater extent, when compared to organizations that are active in production or more simple service concepts.
Grant & Parker (2009) argue that autonomy is thought to stimulate proactivity because it signifies to employees that they have the ability and opportunity to take on broader roles. For example, Parker (1998) argued that autonomy not only directly increases the a persons own control over a task, which boosts self-efficacy, but that autonomy also facilitates enactive mastery experiences by giving employees the opportunity to acquire new skills and master new responsibilities. Indeed, evidence shows that when work is designed to provide autonomy, employees develop higher role-breadth self-efficacy, or confidence in their capabilities to carry out a wider range of tasks and responsibilities effectively. In turn, as a result of their greater self-efficacy, employees tend to set more proactive, challenging goals and then strive to achieve them (Parker S. W., 1997).
2.1.1 Task based autonomy
A determining factor for task based autonomy is equifinality. The concept of equifinality entails that in a lot of different jobs, the end result of the task can be reached via various routes, using different strategies, procedures and also the order in which the whole task is performed
even more true for creative work, because it is difficult to specify a procedure that automatically results in creativeness of an employee. Task based autonomy manifests itself through the changing nature of work in such a way that the individual employee has to be given more authority over a larger area of their responsibilities (Langfred & Rockman, 2016).
Additionally, working in teams in organizations has also contributed to the growth of task-based autonomy. As work was organized in teams more and more, several new constructions based on autonomy emerged, such as self-designing-team, autonomous work teams and self-managing teams (Haas, 2010; Langfred C. W., 2007). This implies that participants of these teams may not experience the autonomy granted to the team in a similar way. “In addition to the greater scope and experience of autonomy, and the pressure on the organization to respond more readily to employee demands and preferences for autonomy, the increasing reliance on team-based work creates many different possible permutations of work arrangements that involve different types of autonomy at different levels” (Langfred & Rockman, 2016).
2.1.2 Job crafting
Another form of autonomy is job crafting. The process of job crafting gives employees the ability to make their own choices in redesigning their jobs to establish a better match between the job and themselves (Tims, 2013). “Job redesign in general has been considered a way to improve productivity and performance for centuries. The Job Characteristics Model is an example, and the tenets of scientific management and Taylorism are clearly rooted in the logic of redesigning jobs to increase efficiency and productivity. Adam Smith’s pin factory example is explicitly about the importance of designing jobs efficiently. What has changed in the past few decades, however, is the entire approach to job redesign. In the past, job redesign has typically been a “top-down” organizational intervention, directed by managers and often in response to technological changes” (Langfred & Rockman, 2016). Nowadays, job redesign can also be a bottom-up initiative of the employee.
The idea of job crafting evolved only recently as a form of individual autonomy and it also shows an increase in speed of the transfer of control from organization to employee (Berg, Dutton, & Wrzenieuwski, 2013; Wrzesniewski, 2001). The employee drives the process of job crafting by shaping and molding the task, by managing and evaluating the personal relations that are important and by having their own cognitive interpretation of the task (Wrzesniewski, 2001). Job crafting is also growing because of the expectations of highly-skilled young employees with regard to autonomy. Drawing from Rousseau (2006), organizations have to take these
expectations of there (future) workforce into consideration in order to stay competitive and be appealing to gifted employees. In PSF’s, the potential for job crafting is much greater, so it has become an increasingly common way that greater autonomy is manifesting itself in such workplaces.
2.1.3 Alternative work arrangements and Idiosyncratic deals
“Another example of how autonomy can take on an expanded role in today’s work is the increasing adoption of alternative work arrangements, including telework, part-time work, flexible work, and the like” (Langfred & Rockman, 2016). Another form of granting choice and authority to individuals stems from self-managed teams, but is extended further as technology enables teams to be spread out physically, but work together in “virtual” teams (Gibson, 2006). Due to increased globalization, work by PSF’s tends to be cross-border with a higher likelihood of virtual teams and autonomous roles.
A specific form of alternative work arrangements exists as idiosyncratic deal. Customizing individual roles in PSF’s has evolved from traditional job design which was focused on top-down interventions and can be viewed as a modern form of autonomy (Hornung, Rousseau, Glaser, Angerer, & Weigl, 2010) In this respect, the idea of task-related idiosyncratic deals was developed, which consists of arrangements that individuals negotiate to create or alter their own job's. “Idiosyncratic deals, in general, are employment terms individuals negotiate for themselves, myriad forms from flexible schedules to career development” (Hornung, Rousseau, Glaser, Angerer, & Weigl, 2010). When creating these i-deals, employees try to gain active involvement in defining the task, Their individual motives are control, self-efficacy and positive social interaction (Bandura, 1997; Ryan & Deci, 2000; Wrzesniewski, 2001). Participating in i-deals asks for personal initiative, which can be defined as the “future-oriented, persistent pursuit of individual and organizational goals (Frese & Fay, 2001). Personal initiative is a form of proactive performance, especially relevant in settings with changing or uncertain job roles (Crant, 2000; Griffin, Neal, & Parker, 2007). Hornung et al. (2010) found that i-deals are positively related to autonomy. The similarity of i-deals with the knowledge work in PSF’s therefore partly supports my hypotheses.
2.1.4 Inhibiting factors for autonomy
Several theories and explanations are offered above of all factors that can be regarded as drivers for more autonomy, since positive results are to be expected when autonomy is granted.
towards formalization and, more formalization means less autonomy (Bailyn, 1985). Adler & Borys (1996) argue that the coercive function of bureaucracy inhibits autonomy, because it inherently strives to more formalization. Another reason is an organizational need for control due to a “negative situation” like financial results that require cost-cutting measures (Dutton, 1987). Organizations give autonomy and then try to control with bureaucratic procedures on the other hand causing employees to work more than they should (Hardy, 2011). Just as there is pressure on many organizations to grant or support autonomy, there is pressure to retain and exert control throughout the organization. Bureaucracy is a powerful and efficient means of controlling people. It is almost an axiomatic assumption that all organizations require a certain level of control to coordinate the actions of their staff members.
2.2 Agency theory and delegation of decision rights
Agency theory poses that people are rational and will make choices based on the choice’s ability to increase either their wealth or leisure. Individuals are assumed to be effort-averse and will only exert themselves when it leads to an outcome with expected utility that exceeds the disutility from the effort. If a task does not add to economic benefit, people will shirk the task (Baiman, 1982).
In addition to the principles of the agency theory, the theoretical model of Abernethy, Bouwens, & van Lent (2004) is useful to research delegation of decision rights, as the granting of autonomy is in essence a form of delegation. The strong analogy between delegation of decision rights and the granting of autonomy can provide some insight in the granting of autonomy.
Decentralization occurs when corporate management assigns decision rights to lower-level managers (Abernethy, Bouwens, & van Lent, 2004). Jensen (2001) argues that corporate management will only decentralize when it can implement a performance measurement system that captures the decision rights allocated to the divisional manager. This condition exists because of the need to assess the contribution of the division to firm value. The reason of the delegation of decision rights lies in information asymmetries between corporate head office and divisions. When better decisions can be made at local level and the transfer cost of knowledge is higher than the cost of control, there is a solid argument for delegation. Ultimately, the costs of control have to be weighed against the knowledge transfer costs as described by Bushman, Indjejikian, & Smith (1995): “Determining the optimal decentralization rights and information in organizations requires balancing the cost of poor decisions due to lack of relevant information against those due to divergent objectives of principal and agents”.
Autonomy is granted for similar reasons: knowledge transfer costs are too high and there is a better understanding of the actual service delivered and local conditions. However, the possibility of installing a performance management system does not exist, since knowledge over the desired outcome and knowledge of the process to achieve the result is not available.
2.3 Self-efficacy and self-determination theory
Social-cognitive or self-efficacy theory proposes self-regulatory cognitive mechanisms that relate to effort. Self-efficacy directly affects effort: a person expends effort on a task that he believes he is able to accomplish. In other words, self-efficacy is thought to be another variable (in addition to goal setting) that affects the key dimensions of effort. Self-efficacy is thought to affect effort through several cognitive, motivational, affective, and task mechanisms. Cognitive and motivational mechanisms include goal setting, expectancies, and the increased use of high-quality problem-solving strategies. Self-efficacy can also have positive effects on initial emotional states (those prior to task performance) and can alleviate aversive emotional states that arise during task performance (Parker S. W., 2006).
Self-efficacy is an ingredient for self-determination, which states that people are motivated better if their needs are fulfilled. Deci & Ryan (2000) recognize the following three essential needs: relatedness, competence and autonomy. Relatedness is the idea of belonging to a social structure. Competence is again related to self-efficacy and is about the person’s own perception of his capability for the task at hand. Autonomy is the perception of being in control of one’s planning and methods to achieve the desired result. Innovative activities cannot be commanded, it is intrinsic motivation that is linked to creativity, which is essential for innovation (Deci, E.L. & Ryan, 2000). If intrinsic motivation is present, that state of mind makes people act upon their natural interests, which leads to innovation (Deci, E.L. & Ryan, 2000). Benner (2003) concluded that “increased process management (which equals less autonomy) was associated with a decline in the number of patents that were based entirely on knowledge new to the firm”. This shows that if innovation is desired, the organization has to grant autonomy to its employees to achieve this result. Self-efficacy and self-determination create the demand for autonomy and should therefore be taken into consideration when implementing controls in a PSF.
2.4 Input controls
Organizations use management control systems to overcome the agency dilemma outlined above in section 2.1 and to ensure that employees actions are in line with the company’s interests (Abernethy & Brownell, 1997). When I focus on bureaucratic mechanisms, three categories of control systems can be identified: (1) behavior control, (2) output control and (3) input controls (Thomson, 1967). Behavior controls regulate actions of employees and require standard procedures and depend on monitoring and evaluations for success. Therefore, the benefits of monitoring have to outweigh the costs. Thomson (1967) argued that “behavior control assumes managerial knowledge of cause/effect relations, also referred to as “task programmability”. Incomplete knowledge by managers of the cause/effect relations hampers the way of defining the specific actions that can be monitored with behavior controls.
Ouchi (1977) stated that output control could be used as an alternative to behavior controls. With output controls, targets are set for employees that measure performance as accurately as possible, while there is freedom within boundaries as to how these targets are achieved. Drawing from agency theory, a risk exists that output control can lead to loss of control, as crucial information is withheld from superiors. To prevent this from happening, elaborate information systems are used to connect appraisal to rewards (Ouchi, 1977). Output control is ex post, which is a disadvantage when it comes to steering and preventing mistakes. Another disadvantage is the possibility of sub-optimization: employees solely focusing on the achievement of results that are rewarded and neglecting important aspects that are not captured by the information system.
Snell (1992) argues that input controls are the only remaining controls that are at the disposal of the manager when knowledge of cause/effects relations are incomplete and standards of desirable performance are ambiguous. In these latter situations, behavior and/or output control could be used. Input controls consist of rigorous selection procedures, training and socialization. PSF’s typically operate in the area (figure 1, right below quadrant) of incomplete knowledge of cause/effect relations and ambiguous standards of desirable performance.
Fig. 1
2.5 Integrativity of Services
The relation PSF’s maintain with their clients shows some typical features. Glückler & Armbrüster (2003) state that the client does not purchase an off-the-shelf product, so he cannot determine the value before committing to a contractual relationship. At the beginning of the project, a contract should be signed that contains all the details of the service provided, but many details of the quality and amount of service are not yet clear. This results in incomplete contracts with regard to the input needed from the client and the output from the service provider (Homburg & Stebel, 2009). The verifiability of the service provided is also difficult, due to the complexity and intangibility, which makes evaluation of the effort troublesome. For verifiability it is necessary that output can be observed, but the intangibility of the service is an obstacle for the definition of objective measures (Homburg & Stebel, 2009). Lastly, the client is strongly involved in the tailoring to his needs of the service provided. Homburg & Stebel (2009) use the term ‘Integrativity of Services’ to qualify the level of customer involvement. “Integrativity means that the agreed-upon service cannot be finalized without the integration of external factors.” This dependence can result in opportunistic behavior from both partners, which needs to be taken in to account when drafting the contract. The amount of integrativity of the service
provided with a number of unknowns from the start therefore bears a relation to the degree of freedom that has to be given to the employee to deliver the service appropriately
2.6 Innovation
Capon et al. (1992) recognize the importance of innovation of products and services for the continuity and long-term success of an organization. Kanter (1983) defines innovation as the process of putting into practice a problem-solving idea. Innovation is the realization, acceptance and implementation of new ideas, processes, products and services. Services providers have placed a strong emphasis on research on innovation (Ostrom, 2010). Hogan (2011) concludes that within the service industry a strong desire exists to understand how organizational innovation is developed in the form of management processes, marketing, design, tailor-made solutions, applying human resources and external collaboration.
The pursuit of innovation by PSF’s is strongly connected to the aspiration to provide services that sprout from the latest technology, best practices and cost efficiency in order to be able to offer advice on the best available and –for the client- best applicable technique to solve the problems the client is confronted with. To be able to realize this, a working environment has to be shaped that stimulates innovative working. In research, the importance of the informal organization in relation to innovation is emphasized. There are conditions for the emergence of innovation. According to Schollhammer (1982), several elements of informal organization seem to contribute to innovation: (1) psychological security and fair rewards for success; (2) continued stimulation and challenge; (3) diffusion of authority and a non-coercive management style; and (4) flexible time and resource schedules. Informal ability to avoid formal barriers to innovation may also be important. Especially items 3 and 4 are part of the definition of autonomy and are cause to expect a positive relation between innovation and autonomy. Besides this, the conception of innovation is a creative process and the creation process also requires more autonomy.
2.7 Task uncertainty
2.7.1 Origin of Task Uncertainty
Task uncertainty is inextricably linked with knowledge work and can be traced back to two main causes. First is the fact that process and outcome become more separated in knowledge work. “Knowledge teams frequently face complex, open-ended tasks for which the a
priori specification of goals and work processes is not possible. Such teams must define these work elements themselves; emphasizing one over the other can lead a team to become either outcome-or process-focused, with implications for the level at which they identify their activities and the flexibility with which they conduct work” (Woolley, 2009). This requires continuous resetting of priorities.
The second cause is a variety of tasks and multiple goals in contemporary jobs for which autonomy is needed to determine their priorities (Langfred & Rockman, 2016). Langfred & Rockman (2016) compare the simple tasks of an assembly line worker with that of a software developer. The software developer’s job has many different aspects, of writing code is just one. The uncertainty of the job is much higher and also involves coordinating with coworkers, planning future projects, testing etc. “As a result of such uncertainty, the autonomy necessary for the knowledge worker typically requires a lot more discretion than that required for more traditional and predictable work in which autonomy may be granted over the specific component of an otherwise fixed process” (Langfred & Rockman, 2016).
An aspect in knowledge work is equifinality, as explained earlier. This effect is a cause for task uncertainty and requires more autonomy and authority to achieve the best performance. Equifinality occurs even stronger in creative tasks, as it is difficult to prescribe a procedure that guarantees creative output.
2.7.2 Task Uncertainty due to: analyzability and many exceptions
Withey et al. (1983) confirm that uncertainty over the outcome of a project is a main source for task uncertainty. Using Perrow’s technology dimensions (figure 2), it is possible to establish that PSF’s score high on both of Perrow’s axes: analyzability and number of exceptions. These aspects result in difficulties to control, continuous target adjustments occur and priorities have to be adjusted requiring more autonomy.
Fig. 2
2.7.3 Task Uncertainty as a result of contracting difficulties in PSF’s
“First, consulting services are often complex and involve a high level of both transactional uncertainty and risk for the client, as he does not purchase a ready-made product” (Glückler & Armbrüster, 2003; Mitchell, 1994). Second, because of the interdependent and interactive character of co-production between consultants and clients in service delivery, both contracting parties can behave opportunistically, which results in a double moral hazard risk that needs to be considered in the design of a contract. Drawing from Homburg & Stebel (2009), Task Uncertainty plays an important part in negotiating contract terms for professional services. The service provider works on an assignment that is difficult to contract, because moral hazard exits for both parties. This is due to the fact that both behavior and outcome cannot easily be measured. So the service provider has to make choices on how much energy to invest in parts of the assignment, which requires autonomy. The autonomy is needed to be able to evaluate and respond to the contribution of the client to the project, to be able to adjust priorities and to negotiate with the client. Third, these contractual relationships are generally rather short-term in nature (Homburg & Stebel, 2009).
3 Hypothesis development
3.1 The relation of Input Controls with Autonomy
Input controls are used, when behavior and output controls are not possible, because monitoring is not efficient and the exact knowledge of the output is not available. Due to the fact that verification of the work is difficult, autonomy has the be granted in this situation. Only by means of socialization can this be constrained to some extent. Abernethy & Brownell (1997) show the relation between input controls and R&D performance in the same situation (see fig 1, quadrant right below).This leads to the conclusion that good performance depends on properly selected employees, that are given the freedom to organize their work and determine priorities and thus I arrive at the first hypothesis.
H1: Input Controls are positively related to Autonomy
The conditions that apply when input controls are used, overlap with several situations where Task Uncertainty occurs. For this reason, I chose not to investigate a possible interaction effect of Task Uncertainty.
3.2 The relation of Integrativity of Services with Autonomy
Integrativity of Services is characteristic for the way service is delivered by the consultant to the client. It is measured by the amount of input and the frequency of synchronization during the process of service delivery. Homburg & Stebel (2009) postulate that fixed-costs contractual agreements that are time and material bound, show a positive link with high integrativity. The client enjoys a high rate of flexibility with regard to a stop/go decision of the project. For the service provider, this induces a strong incentive to produce a high effort, in order to prevent the project of being terminated, even in situations where behavior, nor output are verifiable. It is in the advantage of the service provider to enhance transparency of the service process, in an effort to raise the level of observability of everyone’s behavior. These considerations require sound judgement and tactical adaptations by the executing consultant, which implies: more autonomy. A continuous stream of reporting obligations and consorting with superiors with harm this sometimes delicate process. Therefore, it is my expectation that high Integrativity of Services will result in more autonomy.
The moderator Task Uncertainty amplifies this situation even more, as the added uncertainty forces the consultant to interact even more and to make adjustments even more often, to secure the quality of the final product. One of the causes of Task Uncertainty is the far looser coupling between process and outcomes in knowledge work (Woolley, 2009). This adds to the uncertainty to which the consultant has to react and is cause for the expectation that the larger the Task Uncertainty, the more autonomy is needed to be able to react to this. So, a positive effect of the moderator Task Uncertainty on the relation Integrativity of Services with Autonomy is expected, resulting in the hypothesis:
H3: Task Uncertainty has a positive effect on the relation between Integrativity of Services and Autonomy.
Figure 3: Hypothesis 3
3.3 The relation of Innovation with Autonomy
The striving for innovation can result in the granting of different forms of autonomy. Bailyn (1985) distinguishes between ‘strategic autonomy’ (the freedom to set one’s own research agenda) and ‘operational autonomy’ (the freedom, once a problem has been set, to attack it by means determined by oneself, within given resource constraints) in the setting of a R&D lab, which strives for innovation. This distinction between operational and strategic autonomy can
also be applied to PSF’s. Operational autonomy would be the freedom to interpret a given task, for example: an existing client that is in search of a solution for a problem which is recognized by the consultant as something he dealt with before and for which range of proven concepts exists. Whereas strategic autonomy could be the development of a new service concept, or the modification of an existing solution for a new branch of industry.
The hypothesis that describes the impact of Innovation (defined as the ambition of the organization to be innovative) on Autonomy is derived from the self-determination theory. I expect that the larger the need to be innovative is, the larger the autonomy granted will have to be.
H4: Innovation is positively related to Autonomy.
As strategic autonomy has a much higher task uncertainty, the expectation is that the moderating effect of task uncertainty on the relation innovation with Autonomy will be positive, resulting in the hypothesis:
H5: Task Uncertainty has a positive effect on the relation between innovation and Autonomy.
4 Research method
The research questions have been answered using empirical data that are gathered with an online survey. The survey was created by the “PSF Practice Management Survey Project” of the Amsterdam Business School (Accounting Section) . Joining this research project offers many advantages. First, the targeted respondents are experienced, senior managers of PSF’s who normally would not be easily convinced to dedicate their precious time to a survey of this kind. For one student to assemble the required amount of respondents for a representative sample would be virtually impossible. Second, in this existing project, students gain access to a ready developed survey with meaningful constructs that support the operationalization of the theoretical framework. The sample of respondents consists of 148 partners/managers of PSF’s who are personally contacted via the network of participating students, resulting in a very high response rate. The questions in the survey were established by the UvA PSF project team and are presented in Appendix A. Subject of the study is the management of teams of professionals in PSF’s. This chapter describes the steps take in this research. The first section supplies background information on the selection of respondents and their characteristics. The second section deals with the creation of the constructs that are used in the research. The last section covers the results of some statistical checks and describes the type of tests that were conducted to obtain the results.
4.1 Survey design
The sample is comprised of managers of teams of professionals in PSF’s. Respondents have multiple years of experience in managing (teams of) professionals. An important criterion for selection was their interaction with clients. They were approached via professional networks and family and friends. Students who wished to participate were required to deliver at least six respondents. My own contribution was established at seven respondents. Thanks to the joint efforts of all participating students, a total number of 148 respondents was achieved. The hard requirements for the respondents were: (1) Managing professionals within a PSF, (2) Client interaction as an important aspect of their job, (3) Respondents were stationed in (a) The Netherlands (50), (b) United Kingdom (27) and (c) the rest of the world. The sort of practices were very divers: from consultancy, audit, tax and law to architect, engineers, marketing and design agencies. Besides questions on demographics, 28 questions out of 77 were answered that were used in this study. Missing answers were relatively low: only 10 respondents did not answer
1 or 2 questions. In the statistical analysis the missing questions were left blank. Table 1 shows an overview of the features of the respondents.
Table 1 Characteristics respondents
Sector Consulting Management / Strategy 28%
Accounting 19% Consulting IT 10% Consulting Technology 7% Financial Advice 6% Project Management 4% Software Development 3% Legal 3%
Marketing / Public relations 2%
Other 18%
Practice Size Small 36%
Large 64%
4.2 Survey variables
The questions used in the survey are based on existing surveys previously used for scientific research. The questions are at the base of constructs that are used to test the hypotheses. A construct for the dependent variable Autonomy, a construct for the independent variable Input Controls, a construct for the independent variable Integrativity of Services and a construct for the independent variable Innovation. For every construct is shown which questions are used and what the scientific sources of the questions are.
Before setting up the constructs a check on negative wording was performed, resulting in the reverse coding of the negatively formulated questions. The reliability of the constructs was measured to ascertain to what extent the items that are supposed to measure a single construct, actually measure one and the same thing. The measure for this is Cronbach’s Alpha, which captures inter-item correlation. It has a value between 0 and 1, with higher scores indicating higher levels of reliability. As rule of thumb Cronbach’s Alpha should be at least 0.5 and preferably above 0.7 to have a reliable measure (Field, 2009).
Factor analysis was used to assess convergent and discriminant validity. The convergent validity is established by assessing the communality in item variance from the viewpoint that this variance is in part caused by an unmeasured variable that has an effect on all the items. The discriminant validity has to be established to determine if different items indeed reflect different underlying factors. Furthermore, I can check if all the items really belong to the construct that
they are supposed to reflect and do not fit better with some other construct. The results of the factor analysis are shown in the “rotated component matrix”, leaving out all values < 0.5 and > -0.5 (table 3).
4.2.1 Dependent variable
The questions for the dependent variable Autonomy are drawn from Breaugh (1985) and Hartline & Ferrell (1996). The list of questions consists of the four questions posted below and are measured on a 7-point Likert scale where 1 = strongly disagree and 7 = strongly agree
62. Professionals in this practice may use their own judgment in solving problems 63. Professionals in this practice are encouraged to take initiative
64. Professionals in this practice can schedule their own daily activities
65. Professionals in this practice can choose their own methods and procedures to do their
The Cronbach’s Alpha of the construct Autonomy is 0.664 (table 2), which is higher than the minimum required value of 0.5 and slightly below the preferred value of 0.7. To achieve a higher Cronbach’s Alpha question 65 would have to be left out of the analysis, which would render an value of 0.699. Since this is only a marginal improvement, the original construct was kept and used in the analysis. The component matrix shows that there is one underlying factor that determines 52.3% of the response scores. The results of the rotated component matrix show indeed that this item reflects one underlying factor and that the questions do not correlate stronger with other constructs than this one where it is expected.
4.2.2 Independent variables 4.2.2.1 Construct Input Controls
The questions for the independent variable Input Controls are drawn from (Snell, 1992). The list of questions consists of the three questions posted below and all questions are measured on a 7-point Likert scale where 1 = strongly disagree and 7 = strongly agree.
56. Practice members receive substantial training before assuming project responsibility 57. I have gone through great lengths to establish the best staffing procedures
The Cronbach’s Alpha of the construct Input Controls is 0,823 (table 2), which is higher than the minimum required value of 0.5 and higher than the preferred value of 0.7. There is no possibility of achieving a higher Cronbach’s Alpha by leaving out questions. The component matrix shows that there is one underlying factor that determines 74,1% of the response scores. The results of the rotated component matrix show indeed that this item reflects one underlying factor and that the questions do not correlate stronger with other constructs than this one where it is expected.
4.2.2.2 Construct Integrativity of Services
The questions for the independent variable Integrativity of Services are drawn from Homburg & Stebel (2009). The list of questions consists of the three questions posted below and questions 46 en 47 are measured on a 7-point Likert scale where 1 = strongly disagree and 7 = strongly agree and question 48 is measured on a ratio scale reflecting the frequency of coordination.
46. After agreeing on the service goals, I require little input from the client to deliver the service 47. I are completely dependent on client input in order to deliver the service
48. How often do you typically need to coordinate with client staff during projects?
The Cronbach’s Alpha of the construct Integrativity of Services is 0,568 (table 2), which is higher than the minimum required value of 0.5 and lower than the preferred value of 0.7. There is no possibility of achieving a higher Cronbach’s Alpha by leaving out questions. The component matrix shows that there is one underlying factor that determines 55,9% of the response scores. The results of the rotated component matrix show indeed that this item reflects one underlying factor and that the questions do not correlate stronger with other constructs than this one where it is expected.
4.2.2.3 Construct Innovation
The questions for the independent variable Innovation are drawn from Capon (1992) The list of questions consists of the nine questions posted below and all questions and are measured on a 7-point Likert scale where 1 = strongly disagree and 7 = strongly agree.
2. Our practice is often at the cutting edge of technology
3. The organizational form and structure of this practice encourage entrepreneurial behavior 4. There are special incentives for entrepreneurial behavior
5. Currently I are successful in obtaining talented scientific personnel
6. I develop plans for products and services which span their expected life cycles 7. New ideas are always being tried out here
8. Unusual or exciting plans are encouraged
9. A discussion about the latest scientific inventions would be common here
The Cronbach’s Alpha of the construct Input Controls is 0,844 (table 2), which is higher than the minimum required value of 0.5 and higher than the preferred value of 0.7. There is no possibility of achieving a higher Cronbach’s Alpha by leaving out questions. The component matrix shows that there are two underlying factors that determine 59.4% of the response scores, the first explains 45.3% and the second 14.1%. Analyzing the questions can provide us with a reason of the two factors. Questions 3 and 4 point specifically towards entrepreneurship. This might be a different underlying factor for a different item. The results of the rotated component matrix show indeed that this first item reflects one underlying factor and that the questions do not correlate stronger with other constructs than this one where it is expected.
Table 2 Reliability statistic
Construct Cronbach's Alpha Number of items
Autonomy 0,664 4
Input Controls 0,823 3
Integrativity of services 0,568 3
Innovation 0,844 9
Table 3 Discriminant Validity Rotated Component Matrix1
Component
Construct Survey question 1 2 3 4 5
Autonomy Q62 0,680 Q63 0,731 Q64 0,807 Q65 0,542 Input Controls Q56 0,845 Q57 0,758 Q58 0,827 Integrativity of Services Q46 0,696 Q47 0,720 Q48 0,771 Innovation Q1 0,559 Q2 0,558 Q3 0,808 Q4 0,782 Q5 0,574 Q6 0,768 Q7 0,769 Q8 0,604 Q9 0,733 4.2.3 Interaction variable
The questions for the interaction variable Task Uncertainty are drawn from Gresov (1989) and Withey et al. (1983) the list of questions consists of the nine questions posted below and all questions and are measured on a 7-point Likert scale where 1 = strongly disagree and 7 = strongly agree.
33. In our practice, problems for which there are no obvious solutions arise frequently 34. Tasks and activities in our practice do not change much from day to day 35. Activities in our practice are repetitive
36. Activities in our practice are routine
1 Extraction method: Principal Component Analysis, Rotation Method: Varimax with Kaizer Normalization.
37. Practice members do the same job in the same way from day to day and client to client 38. There is an understandable sequence of steps to perform practice activities
39. Practice members can rely on established procedures and practices to perform their activities 40. There is a clearly known way to do the major types of work in this practice
41. How often do issues arise that require different methods and approaches?
The Cronbach’s Alpha of the construct Task Uncertainty is 0,819 (table 2), which is higher than the minimum required value of 0.5 and higher than the preferred value of 0.7. There is no possibility of achieving a higher Cronbach’s Alpha by leaving out questions. The results of the rotated component matrix show indeed that this first item reflects one underlying factor and that the questions do not correlate stronger with other constructs than this one where it is expected.
4.2.4 Control variables
This study uses Small Practice and Consulting as control variables in an effort to increase the accuracy of the tested relations between independent and dependent variables. One of the demographics questions was the size of the practice managed by the respondent. The assumption is that a small practice could influence the end result, since autonomy is closely related to span of control. In a small practice with more frequent contact and specific knowledge of the attitude and capabilities of an employee, it might be easier to grant autonomy than in a large practice, where formal controls might be installed. To see if a difference exists between small practices and large, a dummy variable was created, where (1) = Small Practice (1 – 49) and (2) = Large practice (> 50).
Second, I have divided the results of the sort of practice into two categories: Consulting and Other, also by creating a dummy variable. It is my expectation that in consulting firms, the demand for autonomy might even be bigger, as Task Uncertainty is an inherent feature of the work.
4.3 Statistical model
analysis. The first step in the analysis are the descriptive statistics, as shown in table 4. This analysis shows that the 22% of the respondents works in a Dutch-based company. Furthermore, the data show that 47% is working in a Consulting firm. The scores on variables Autonomy, Input Controls, Integrativity of Services, Innovation and Task Uncertainty are relatively high. Table 4
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation Small_Practice 147 0,00 1,00 0,3605 0,48180 Consulting 148 0,00 1,00 0,4700 0,50100 Autonomy 148 3,25 7,00 5,6481 0,76918 Input_Controls 148 1,00 7,00 4,7432 1,40138 Integrativity_of_services 148 2,50 6,00 4,0225 0,45908 Innovation 148 1,33 6,56 4,6577 1,04365 Task_Uncertainty 147 1,22 6,22 3,9184 0,96921
The correlation between the variables is computed and presented in table 5. A significant correlation is reported between Innovation and Autonomy, however, the other independent variables Input Controls and Integrativity of Services do not show a significant correlation. Table 5
Auto
nomy Input Controls Integrativity_of services Innova tion Task_Un certain ty Small Practi ce Consul ting
Autonomy Pearson Correlation Sig. (2-tailed) Spearman's rho Sig. (2-tailed)
Input_Controls Pearson Correlation 0,158
Sig. (2-tailed) 0,055
Spearman's rho 0,121
Sig. (2-tailed) 0,142
Integrativity_of_s
ervices Pearson Correlation -0,004 0,026
Sig. (2-tailed) 0,963 0,753
Spearman's rho 0,025 0,068
Sig. (2-tailed) 0,762 0,409
Innovation Pearson Correlation 0,271** 0,436** -0,135
Sig. (2-tailed) 0,001 0,000 0,101
Spearman's rho 0,257** 0,375** -0,090
Sig. (2-tailed) 0,002 0,000 0,276
Task_Uncertainty Pearson Correlation 0,329** 0,001 0,129 ,166*
Sig. (2-tailed) 0,000 0,993 0,120 0,044
Spearman's rho 0,350** -0,003 0,106 0,154
Sig. (2-tailed) 0,000 0,970 0,201 0,063
Small_Practice Pearson Correlation 0,133 -0,222** -0,202* 0,019 0,098
Sig. (2-tailed) 0,109 0,007 0,014 0,820 0,238
Spearman's rho 0,125 -0,236** -0,210* -0,001 0,096
Sig. (2-tailed) 0,131 0,004 0,011 0,990 0,247
Consulting Pearson Correlation -0,006 -0,058 0,141 0,051 0,144 0,082 -Sig. (2-tailed) 0,938 0,481 0,088 0,537 0,081 0,325 Spearman's rho 0,000 -0,090 0,169* 0,054 0,182* 0,082 -Sig. (2-tailed) 0,997 0,275 0,040 0,518 0,028 0,325 ** Correlation is significant at the 0.01 level (2-tailed).
4.3.1 Equations
The hypotheses H1, H2 and H4 have a cause and effect relation. These hypotheses predict that the variance in the dependent variable is caused by an independent variable. The measurement level is ordinal. The following equations are executed using a multiple regression analysis.
Model I: Autonomy = 𝛼 + control variables + error
Model II: Autonomy= 𝛼 + Input Controls + Integrativity of Services + Innovation + control variables +
error
Hypotheses H3 and H5 are hypotheses with an interaction variable. These hypotheses predict that the variance of an independent variable on a dependent variable is caused by a third variable. The level of measurement is non-categorical. For these hypotheses as well, a multiple regression analysis was used. For the calculation of the interaction variable, Task Uncertainty is multiplied by Integrativity of Services or Innovation. Since this can cause multi-collinearity problems, centered variables were used to avoid this. As a result of the use of mean-centered variables, the standardized bèta’s lose their meaning, therefore, the unstandardized bèta’s are reported in the results.
Model III: Autonomy= 𝛼 + Input Controls + Integrativity of Services + Innovation + Task Uncertainty +
Integrativity of Services x Task Uncertainty + control variables + error
Model IV: Autonomy= 𝛼 + Input Controls + Integrativity of Services + Innovation+ Task Uncertainty +
Innovation x Task Uncertainty + control variables + error
Model V: Autonomy= 𝛼 + Input Controls + Integrativity of Services + Innovation+ Task Uncertainty +
5 Results
This chapter contains the results of the regression analysis. Table 6 shows the results of the analysis with Autonomy as the dependent variable. Model I is a baseline model in which only the control variables are tested. These control variables are taken into consideration in the following models. Model II tests the direct relation between the independent variables and the dependent variable. Model III expands Model II by including the direct effect of the interaction variable Task Uncertainty and the interaction effect of Task Uncertainty on the relation Integrativity of Services with Autonomy. Model IV expands Model II by including the direct effect of the interaction variable Task Uncertainty and the interaction effect of Task Uncertainty on the relation Innovation with Autonomy Model V tests the effects of all independent variables and both interaction variables together in one model. The R-squared indicates which part of the variance in the dependent variable is explained by the independent variables of every model. The ANOVA further produces the F-statistic, while the P-value (sig) indicates whether or not the model is significant.
Dependent variable Table 6
Autonomy
Model I Model II Model III Model IV Model V
Control variables Small Practice 0,133 0,158 0,127 0,129 0,131
Sector (Consulting) 0,001 -0,100 -0,041 -0,047 -0,046
Independent variables Input controls 0,085 0,105 0,105 0,106
Integrativity of services 0,058 0,007 0,008 0,007
Innovation 0,238 0,178** 0,166* 0,167*
Task Uncertainty 0,292*** 0,300*** 0,298***
Interaction variables Task Uncertainty * Integrativity of services 0,029 0,021 Task Uncertainty * Innovation -0,025 -0,023 R2 0,018 0,098 0,179 0,179 0,180 R2 adjusted 0,004 0,066 0,137 0,138 0,132 F-value 1,289 3,069** 4,292*** 4,312*** 3,751*** P-sig 0,279 0,012 0,000 0,000 0,001 * significant by P < 0.1 ** significant by P < 0.05 *** significant by P < 0.01
5.1 Control variables
Model I tests the control variables Small Practice and Consulting. The ANOVA table shows that the model is not significant (P = 0.279). This implies that both variables do not impact the tested relations in the main model.
5.2 Main model: test of Input Controls, Integrativity of Services, Innovation and Autonomy
Model II has an R-squared (adjusted) of 9.8%, so model II explains 9.8% of the variance in the dependent variable. The ANOVA analysis shows that the complete model is significant (P = 0.012). A positive relation between Input Controls and Autonomy was expected (H1). The test results of model II do not support this as has a beta of 0.085, but it is not significant (P = 0.358), so H1 has to be rejected. Furthermore, a positive relation was expected between Integrativity of Services and Autonomy (H2). The test results of model II do not support this as has a beta of 0.058, but it is not significant (P = 0.484), so H2 has to be rejected. Finally, a positive relation was expected between Innovation and Autonomy (H3). The test results of model II do support this with a bèta of 0.238 and a P-value of 0.010.
5.3 Interaction effect of Task Uncertainty
Model III has an R-squared (adjusted) of 13.2%, so model III explains 13.2% of the variance in the dependent variable. The ANOVA analysis shows that the complete model is significant (P = 0.000). The expectation of the interaction effect was that Task Uncertainty would have a positive effect on both the relation Integrativity of Services and Autonomy and on the relation Innovation and Autonomy. Model III tests the interaction effect of Task Uncertainty on Integrativity of Services. The test results of model III do not support this as has a beta of 0.029, but it is not significant (P = 0.810), so H3 has to be rejected.
Model IV has an R-squared (adjusted) of 13.8%, so model IV explains 13.8% of the variance in the dependent variable. The ANOVA analysis shows that the complete model is significant (P = 0.000) Model IV tests the interaction effect of Task Uncertainty on Innovation. The test results of model IV do not support this as has a beta of -0.025, but it is not significant (P = 0.679), so H5 has to be rejected.
Model V has an R-squared (adjusted) of 13.2%, so model V explains 13.2% of the variance in the dependent variable. The ANOVA analysis shows that the complete model is significant (P = 0.001) Model V tests both the interaction effect of Task Uncertainty on Integrativity of
Services and of Task Uncertainty on Innovation. The test results of model V do not support this, both values differ slightly from model III and IV.
5.4 Robustness test
The component matrix (table 5) of the construct Innovation reflects two underlying components. Analyzing the questions can provide us with a reason for the existence of two factors. Questions (3) “The organizational form and structure of this practice encourage entrepreneurial behavior” and (4) “There are special incentives for entrepreneurial behavior” point specifically towards entrepreneurship. It might be that entrepreneurship has a different association and loads on a different factor.
To understand the difference that these two survey questions could have on the results of the statistical analysis, a new analysis was performed without these two questions. The newly formed construct Innovation still has a high Cronbach’s Alpha of 0.837. The results of the regression analysis can be found in table 7. Leaving out the two questions has some impact on the results: beta’s change slightly and significance drops. Only in model III, Innovation is significant.
Table 7 Dependent variable Autonomy
Model I Model II Model III Model IV Model V Control
variables Small Practice 0,133 0,168 0,133 0,141 0,141
Sector (Consulting) 0,001 -0,004 -0,038 -0,050 -0,049
Independent
variables Input controls 0,117 0,130 0,133 0,134
Integrativity of services 0,052 0,000 0,004 0,003 Innovation 0,179* 0,129 0,112 0,112 Task Uncertainty 0,303*** ,314*** 0,313*** Interaction variables
Task Uncertainty Centered * Integrativity of services
Centered 0,026 0,011
Task Uncertainty Centered *
Innovation Centered -0,043 -0,043 R2 0,018 0,080 0,168 0,171 0,171 R2 adjusted 0,004 0,047 0,126 0,129 0,123 F-value 1,289 2,44** 3,981*** 4,075*** 3,541*** P-sig 0,279 0,037 0,001 0,000 0,001 * significant by P < 0.1 ** significant by P < 0.05 *** significant by P < 0.01
6 Conclusion
This study seeks to contribute to MCS research in the following way. The three research questions are aimed at exploring what the determinants of autonomy are, when faced with Task Uncertainty arising from various sources. Evidence from 148 managers employed by PSFs suggests that Input Controls and Integrativity of Services are not in general associated with autonomy, while Innovation can be linked to Autonomy. The moderating effect of Task Uncertainty could also not be established.
6.1 Discussion of findings
The relation between input controls and autonomy (H1) that has been researched is based on the expectation that the use of input controls would result in more autonomy. In situations, where input controls are the only remaining control possibilities, I expected that the only logical result would be that employees would demand and receive more autonomy. The idea that with very stringent selection and training procedures installed, organizations would proceed and implement a controlling environment, seems contradictory. Since input controls are also linked to performance in R&D environments, and performance requires autonomy (Abernethy & Brownell, 1997), a relation could be expected. The regression analysis however, shows a positive relation, but it is not significant. So, in this research setting, the direct effect could not be proved. The low R-squared (adj) of 0.047 indicates the small explanatory power of the model. A possible explanation could be the effect of the organizational culture within PSF’s. Based on Henri’s (2006) organizational cultural types, a flexible culture that advocates loose and informal controls would be expected be in PSF’s. But if in fact a more rational and/or hierarchical culture dominates, this could certainly have its effect on Autonomy.
Next, I examined the relation between Integrativity of Services and Autonomy (H2), for which a positive effect was expected as well. The expected effect is based on the way the delicate relation between service provider and client has to be managed. Fixed fee contractual terms occur more often in combination with high Integrativity of Services and with this sort of contract, a moral hazard occurs (Homburg & Stebel, 2009). The service provider has to manage this relation, for which more autonomy is needed. For this effect also, no significant result from the regression analysis can be reported. A possible explanation is the relatively low Cronbach’s Alpha of 0.568 for this construct, this value is above the minimum required, but below the preferred value. So the construct may not be fully reliable and thus, the questions may not fully
which indicates the small explanatory power of the model. A possible explanation for this lack of evidence might be the existence of bureaucratic tendencies that impose rules, roles and standards. These oppose the proposed self-supervision -which can be compared to Autonomy- by Mills (1983) in PSF’s.
Thirdly, the interaction effect of Task Uncertainty on the relation Integrativity of Services and Autonomy (H3) was studied. The expectation that this would be a positive relation comes from the idea that uncertainty makes the interaction with the client more complex. More coordination with the client is needed to safeguard an end-product that is up to standard, but at the same time, the uncertainty can cause overspending the original amount of time planned for the project. The expected effect however, cannot be confirmed by the regression analysis. Besides the lack of significance, the coefficient is very low, so even if it could be established, the effect would be neglectable.
The fourth relation between Innovation and Autonomy (H4) was based on the theory that if an Innovative approach was the ambition of a PSF, this would lead to a policy where more Autonomy was granted. The self-determination theory (Deci, E.L. & Ryan, 2000) poses that for an intrinsic motivation, which is required to achieve innovation, autonomy is needed. The regression analysis confirms this hypothesis.
The fifth relation is the interaction effect of Task Uncertainty on the relation Innovation and Autonomy (H5). As different levels of Innovation –operational or strategic- are connected to more uncertainty, the expected effect was positive. The effect however, cannot be confirmed by the regression analysis. Besides the lack of significance, the coefficient is very low and negative, so even if it could be established, the effect would be neglectable.
6.2 Limitations
Similar to most studies, there are limitations to these findings. First, and probably most importantly, this paper investigates the newly developed constructs of Input Controls, Integrativity of Services, Innovation, Task Uncertainty and Autonomy. In particular the construct of Integrativity of Services shows a low Cronbach´s Alpha of 0,568, which may cast doubt on the validity.
Second. the specifics of the research project may be the origin of some limitations. The most important is the business-related terminology used in the survey questions. The concepts might very well be interpreted differently across categories of organizations, because of the different meaning that they have in their surroundings. Since the survey was set up in English
and respondents were mainly non-native English speakers, small interpretation differences might occur that have resulted in biased responses (Harkness, 2012).
6.3 Implications for future research
MCS design has been researched by many scholars, also with a focus on PSF´s. The role of autonomy as design choice has been clearly marked in this paper, however, of the direct relations that were researched, only the determinant Innovation could be established. As discussed in the discussion on Integrativity of Services, the effect of bureaucracy could have a decisive impact on autonomy. This could be a direct, but also a moderating effect on the relations in this paper.
